Good morning, everybody from San Diego. We want to welcome you all back to your streaming broadcast coming to you directly from the world of solutions here at Cisco Live 2025. So exciting. I'm Steve Multer on behalf of the entire Cisco TV team. We are really delighted to have you with us today. Now, from the moment this event kicked off, we have been experiencing the story of all of the unique, all of the dynamic ways that Cisco is powering how people and technology work together across the physical and the digital worlds. All of the different ways that Cisco is leading the revolution in how infrastructure and Data Connect and protect organizations exactly like yours, like all of ours in this constantly shifting, advancing AI era. The fact is, this is all about outcomes, right?
Customer value, truly secure networks, complete end-to-end observability, unparalleled collaboration, everything that you and that your organization need to thrive and to excel thanks to the depth and the breadth of Cisco's combined portfolio of technology and solutions. It's been an amazing, amazing week already, and it just gets better and better today. Right now, we're about 28 minutes from our Wednesday morning keynote kicking off at 8:30 A.M. The title, Innovation in Action. This is where the big picture rubber hits the operational and performance road. We are going to get detailed stories from Chuck Robbins, from Carrie Palin, Liz Centoni, Oliver Tuszik, along with some great special guest partners on how Cisco is harnessing the power of AI, driving growth, helping to build a better world for all of us. We're going to hear details on the innovations that we have created here in the AI era.
We'll take a specific focus on Customer Experience, great use cases of how Cisco is helping you and helping your organization to navigate all of it. It kicks off, like I said, right at 8:30 A.M., so stay right here with us. Also, make sure you keep hitting us up on social media, whatever platform you like. Great. Go ahead and post your comments there, your thoughts, your ideas, your inspirations that you're hearing. Maybe you put a photo in there, maybe put up a video. Just make sure that you also include #CiscoLive and @CiscoLive so our social media team here at the event can see those posts and respond. I'm here in the studio with another one of your five wonderful Cisco TV hosts. Good morning, Lauren White.
Good morning. Good morning, Steve. How are you today?
This is just, you know, the energy, it sort of builds throughout the week. You get on site, Lauren, and it's a little overwhelming and it's chaotic and you're running around and trying to figure out what goes where. As each session passes and each time you walk around and see something new, the excitement just builds and we get more confident. I'm going to go way out on a limb and say we get a little bit more proud. Am I allowed to say that of being part of the Cisco ecosystem?
Yes. No, I completely agree. The best part about it is that we're all here together. Nobody's alone. If you have a question, you can either ask someone sitting next to you or go over to the World of Solutions and ask the actual product the same question. As a customer, I don't think there's any better thing you could want to do, right? Like all those questions you've been itching to ask, you have access here at Cisco Live.
100%, Lauren. We talk about community all the time. Community is more than just getting together with friends and hanging out in the social lounge and having a coffee together. Community means, as you just said, getting into the space and meeting directly one-on-one with an engineer and saying, "I've got this challenge that I need to overcome. I've got this obstacle." I know Cisco tells me that you can apply AI to create the data center that I actually need or to be able to drive my network and build its power. How do I actually get it done and how do I leverage the investments I've already made with Cisco? Where else can you do that one-on-one with an engineer and set up that space, right, Lauren?
Exactly. I don't think you can do it anywhere else. This is the place to do it. We have this chance every year. We are always excited to see you all here and taking advantage of who you have access. If you see me, ask me a question because I'll point you in the right direction. Like you have access to all of us here.
We are going to find Lauren because Lauren's got enough energy for all of us. You will see her out there on the concourse. Speaking of fantastic energy, my dear friend Robb Boyd is out in the keynote being our eyes and ears out in that amazing space. Good morning, my friend.
Good morning, sir. Yes, I have my eyes open. I was just watching and I was commenting because what's interesting is when you get in here early, as we get special access because we're very special people as a part of the broadcast, is watching the doors first open and then watching the nerd walk. I mean that with all positivity. I do not mean any shame because I do the same thing. It is like, let's get a good seat. You are in and you are scanning and you are figuring out, okay, there are some green things that just say reserved. Over here, there are some red NetVet slash CCIE. There is another reserve section that may or may not be that well marked. Everybody is figuring it out. There are these guys, if you look to your left, Julie, IT leadership.
See, that was a sign I was looking for earlier. These are all IT leadership. Let's salute them, thank them for their service. Appreciate them leading our IT. We need more IT leadership. Thank God they're all flowing in so we know that we're in good hands. This is going to be so much fun. Steve, you and I have talked about this before, but the idea is that there's always kind of a turning point from that first day jitters of let's get the big announcements out so that we can release the press releases and get the backup information that we were struggling to then kind of connect with what we saw on stage yesterday.
I would assume that what we're in store for today on stage is going to be more detail, more demonstrations, and more of the second level of stuff that we all need, both from the executives that are speaking as well as the information that is being shared. We'll start to get a lot more detail about how we're actually going to implement what stuff is actually real and shipping versus being announced as coming up and still needing input, perhaps. Either way, this has been a fantastic crowd here. It's exciting to see them come in. I'm just going to be digging up a few. I'll be digging up a few interviews. I don't know exactly who they're going to be yet. Got a few potentials. We'll see if they make the cut. Either way, I'll throw back to you guys in the studio.
If you promise to check back with me and my other co-hosts here in the keynote.
That is a promise you can count on, Robb. Thank you again for bringing us the action out there. We do appreciate it. Actually, what Robb just said, I think that's so important, my friend, because yes, yesterday is more conceptual. What are we doing? What are the new releases? What are the new announcements that we're making here at Cisco? Today is where we take those new announcements and we actually make them applicable. Everything conceptual is great. It gets us excited. It gets us inspired.
Then how do we actually get things in action, start to apply them, figure out what works best for us so that we can be specific for our own needs, whether, again, we're somewhere on the data center journey or we're on that journey to the cloud or we're looking to be able to add additional power into the network to be able to support these new AI demands. It's an extraordinary quantity of bandwidth and power that's required. You know what you need for your organization. Here is where you get the answers that you are looking for. It starts today. We're going to go back out to the show floor where my friend Michelle Moreira is out there. Good morning.
Hi, yeah.
Hi, guys. Hello. I'm so excited to be here on day two. Z, how are you feeling?
I am super excited, Michelle. I tell you, you know what kept me up last night is my passion to fuel your appetite to be here in the years to come. I tell you, the momentum is building in this room. I wish we had Smell-O-Vision, Phil-O-Vision, but you know what? I just read that you know what's on the agenda today? I do not think we're far off. They're going to be talking about quantum revolution. It's closer than you think. You want to tune into that today? I tell you, you need to be here. If you're not at Cisco Live, I do not know where you are. Wherever you're tuning in from your living room, your car, wherever you are, do not just watch. You need to experience this. This is an experience. React, connect, tag us on social media. How about you, Michelle? How are you feeling?
She just got me even more excited. All right, today's keynote, Innovation in Action. We're talking about AI, all things AI today. If you guys are looking at your description on your phone of all of the breakout sessions, you can see the keynote. They have a description that says AI at Cisco is going to solve a lot of the world's pobblems. How are we going to land that message today? That's what I'm excited to hear.
We're doing it. We're doing it, Michelle.
All right. Back to you, Steve, in the studio.
Thank you so much. I appreciate it. Z, Michelle, I can't wait to see you guys a little bit later. We're always so separated. I'm up here in the studio. They're out there having fun out in the keynote space. I just look at them and I see their smiles and their great conversations that they're having and bringing the excitement to us here on the show floor. When we talk about what is available here at the show, we always have this conversation. Why be here in person? We've got this phenomenal broadcast that comes out. Hey guys, if you could bring down the guests over on the other side. I can't hear it. There we go. Now we're back with each other. This is the beauty of live theater, everybody.
When we think about what we actually see out here on the show floor, it takes the opportunity to walk around, to connect with people, to get a sense of the depth and the breadth of the Cisco portfolio. A lot of times we focus in on our one thing. I'm a security guy. I work in networking. I want to see what's going on in the data center. I want to stop into the NOC and check out what's going on there. It's when you're here that you get to walk in, you get that full big picture. It gets really exciting. We are going to go, I'm sorry, where are we going? I missed that. Okay, great. We're going to go over to Lauren. Lauren is here with some special guests. I appreciate it. It's great. This is, again, this is the excitement of the morning. We don't know what's happening. It all flies in. Lauren, take it away.
Hello, hello, everyone. I am here with two amazing guests, Kevin Kennedy, VP of Product Marketing for the Security unit here at Cisco, and also Matt Caulfield, VP of Product Duo and Identity Security here at Cisco. Thank you both for joining me today.
Good morning.
Kevin, I'm going to start with you with this first question, if you don't mind, and we jump right into it. How would you describe Cisco's Zero Trust Network Access strategy and like what we're doing in that space, and how is it related to SASE?
Sure. Great question. We are redefining it to make Zero Trust Access truly universal, and we are eliminating blind trust. Zero Trust Network Access came of age during the pandemic, and it was designed to solve the problem of that time, which was connecting employees at home on laptops to what turned out to be some set of applications. That is not enough today. We are taking it further and securely connecting everyone, whether it is employees, contractors, partners, IoT devices, soon agents, to any application from a legacy application to an AI resource from anywhere, home, office, even distant locations like oil rigs and airplanes where you have bad networks. We do the plumbing. Users log in and get to work. They have a consistent experience, and security teams have a consistent policy. For Cisco, we went further. What do people expect from the network?
It just works. We measure the experience. We give the tools to troubleshoot any issues, and we even predictively say, hey, something's going to go wrong and stop it before it happens, both for the network and for the policy. We have built this all to the SASE point. We have built this all on the foundation of single vendor SASE from market leading SD-WAN to modern SSE, Thousand Eyes, VPN. There is a lot of technology that we have built over the years that is baked in. For customers who have those technologies, it makes it even easier to get going. There is the eliminating blind trust part.
That was a good segue. I think that was a perfect segue. This one, Matt, is for you. How would you talk about what our identity-first strategy looks like?
Yeah, of course. Not many people think of Cisco as an identity company, but we are. We do a ton of business in identity. Recently, what I'm most excited about is that Duo launched Duo Identity Access Management just a couple of weeks ago. Last week, I was at Identiverse, which is another big conference for identity folks. This week, we're here. We launched Duo from what people know it as, which is just multi-factor authentication. You get a push notification to your phone. You press allow into a full identity suite. Identity access management gives companies, Cisco customers, the ability to use Duo as their standalone identity provider or to use it with their existing identity provider. There's a lot we can do there. What makes this also great is what Kevin was talking about.
We tie this into our Zero Trust strategy and our universal Zero Trust architecture. Identity is a huge differentiator. Identity Intelligence, Identity Access Management is a huge differentiator. We're one of the only vendors for Zero Trust in the industry that has an integrated identity solution.
Wow, wow. I have to ask, right, because we've been hearing a lot about AI and Agentic AI here at the conference. How does the solution really fit into that picture?
Yeah. One of the big topics right now is Agentic AI. How do I, as a—I'll give you an example. As an employee, spin up a bunch of agents to do some work for me. I can 10x my productivity if I only had like an email inbox agent and a travel booking agent and somebody to do market research for me. All of this has been impossible up until now because agents and AI have been stuck in a sandbox with read-only data. Maybe you have a RAG API but that's about it. We're going to unlock that with universal Zero Trust by giving agents the same access as employees have to resources, SaaS, Cloud, on-prem resources, so they can start taking actions on our behalf.
Cisco, through our Zero Trust architecture, through agentic identity, through agentic access, we're going to make that possible in a way that's safe and secure for our customers.
Amazing. I'm excited. I just love the fact that we are doing this with that one Cisco in mind too. Like in this whole conversation, I hear the collaborative pieces and how it all fits together like a puzzle. This has been amazing. It's been a pleasure speaking to the both of you. I am going to pass it back over to Steve.
Lauren, so appreciated. Kevin and Matt, really great to have you with us. Thank you for taking the time to be in studio. Great to get that ZTNA inside story. Zero Trust is such an important and powerful part of advancing the network of onboarding AI capabilities. When we look at being able to do AI responsible, a huge part of that is security has got to be embedded from the core. Really appreciate the two of you being here. We are going to head directly back out into the keynote space where Z is getting a little quality time with Rodney Clark. Hey there, Z.
Hi there, Steve. I'm here with Rodney Clark, our Head of Partners. Rodney, tell me this. You know, there's so many exciting announcements going on, innovations that we're releasing. You know, you're responsible for all of our partners. What really keeps you up at night or what's your passion and what do you want our partners to get out of this week and what really set your fire, set you aflame?
For our partners, just like all of us here at Cisco, this is like the biggest selling event of the year.
Yeah.
We have over 150 partners participating in this in the world of solutions. They're hosting their customers. We're participating and hosting alongside to try and accelerate business. The cool thing that my partners are telling me is yesterday and the innovation and G2's book of announcements has been amazing. Three things that they're excited about: Unified Catalyst, Meraki platform, all that coming together, agentic ops and AI Canvas. Oh my gosh, an amazing services opportunity for our partners. Our ongoing development on pods, AI pods specifically, that Splunk pod announcement was pretty cool. They are fired up and they're using that energy to engage with their customers, trying to accelerate opportunities.
Wow. Rodney, I am super excited for what's next. I am sure you're going to be reporting back to us all the things that you're hearing. Steve, back to you.
Thank you so much, Z. I appreciate it. Please give our thanks to Rodney as well. He's just an amazing voice for Cisco, great advocate, remarkable mind, and great passion and heart for the brand and for what it is that we do and what it is that we bring to the marketplace. We are about 12 and a half minutes out from the day's opening keynote. Stay with us here on the broadcast. Make sure to keep reaching out to us using social media. Just add in #CiscoLive and @CiscoLive so our social media team can see and respond to all of your posts. Thank you for engaging with us throughout the show. I'm super excited at this point. I'm looking up in the monitor that's like slightly over to my left. And what do you know? There's Michelle hanging out with rock star superstar Carrie Palin. Hey there, Michelle.
Hi guys. Welcome back to the show flow. I am here with Carrie Palin, SVP and Chief Marketing Officer at Cisco. Carrie, I'm such a fan. I can't believe I get to interview you. Today, I'm going to ask you a couple of questions. The first is innovation in action. What can we be looking forward to, thinking about, tuning into during this keynote?
First of all, yesterday, we dropped a massive payload on innovation. It was just announcement after announcement after announcement. Everybody was in awe with all of the amazing innovation coming out of the portfolio right now, and especially cross portfolio, which is super exciting. Today, we're going to hear from customers, Ford and University of California, San Diego, talking about how that innovation is actually being put to action to drive their outcomes. That's my favorite part, which is really we can talk all day long about the amazing tech we're bringing to life, but seeing it move the needle for our customers through their voice, money.
That's powerful.
Yeah, it's powerful.
All right. I have to ask one more on AI. In this era of AI, how can we be mindful of our Customer Experience?
Oh, yeah. I was actually talking with an analyst yesterday at our analyst roundtable. I had this comment. He said, yeah, I actually get that. I was like, look, AI is amazing because it's so much it's helping us just supercharge IQ. What AI doesn't have is EQ. We have to drive a level of intimacy with our customers where we not only know them, but we anticipate how they're feeling and when they need us and when they need us to back off. That level of nuance is not AI. That nuance is human beings that are wrapping that experience around the AI and doing it together. That will never change.
It is so important that we have all these big, beautiful brains in the room with us today, all these lovely humans, because they're the ones who are going to be doing that out in the ether with their customers.
Absolutely. Carrie, thank you so much for your time. I'm excited to hear you talk and everybody talk today at the keynote and throughout the day. Back to you, Steve, in the audience.
Thank you so much, Michelle. I think Carrie may have just hit my favorite thing of the morning. Big, beautiful brains. That's exactly what this event is all about. It's about the Customer Experience, as Carrie also just said, right? If we are not creating better outcomes, then what is the point of the technology, the service, the solution that we are able to build? We are a phenomenal organization. We have the best partners, the best customers in the world, but it's about getting to better outcomes for the individual because what are our companies, our organizations made up of? Tons and tons and tons of individual humans, each with their own specific need, their own demand, their own vision of what success looks like. What Carrie just said is so brilliant. How do we listen enough and then back off enough to give them their autonomy?
As Chuck said yesterday at the top of the opening keynote on Tuesday, he said that whether we like it or not, the good news and the bad news, it all comes down to us individually. How do we go to our leadership? How do we go to our teams? How do we go to the people making the executive decisions and say, here is what we need in order to thrive, in order to create success? It's all on us, the individual. We are the users with the user experience as much as the people that we're selling products to or partnering with. Really excited to hear from Carrie. She's just, again, a remarkable leader for us here at Cisco. I am again glancing off to my left using my peripheral vision right over here. I see Robb Boyd back out there in the keynote space with another special guest. Robb, let me send it out to you, my friend.
Yeah, absolutely. I made a new friend here from Indianapolis, correct?
That is correct.
I was curious. Oh, there are our hosts going by behind us. Pay no attention to behind the scenes here. Jim, you were telling me, I know you specialize. You had built a company up, specialized against Splunk Technologies.
Yeah.
You guys were acquired by Presidio?
Yeah.
What I really wanted to find out from you, I feel like Splunk, no secret that Splunk has been a great company to work with because Cisco had been working with them long before the acquisition. I feel like we're seeing more integration now. I'm just curious about the effective working with more customers.
Yes.
What do you think the greatest strength is? Do you feel like we're hitting it?
Hey, look, my team, we've been working with Splunk for like 13 years.
Yeah.
I now lead the Splunk practice for Presidio. We're a dedicated Cisco partner. Look, like everyone at Cisco, they won the lottery, okay?
Yeah.
Okay? Like a lot of people think that Splunk is a security tool. It's like, no, no, no, no, no. It's a data platform.
Okay.
Okay? Like we're going to learn a little bit later today, like powering the AI data center. Like, what do you need for that?
Yeah.
You need data.
Yeah.
Splunk is the Mac daddy in the marketplace for collecting, indexing, and allowing analytical insights out of data.
Sometimes I think we look at the observability. We look at the ability to see deeper into what is actually happening in the entire stack, right?
Right. Yep.
That lends itself to security as well. It lends itself to us.
Oh, totally.
Understanding, but understanding is required to make decisions, especially when it comes to how do we move to automating these things and having trust in it, right?
Yes. Yeah. We heard yesterday, you know, security and observability coming together like this.
Yep.
The future is Agentic AI. What does Agentic AI need? It needs data. That is why we saw like in one of the slides that digital resilience thing on the bottom.
Absolutely.
Like people are like, well, what the heck does that mean?
Yeah.
It's like, we're going to get resilience on the infrastructure through data.
You're right.
Splunk makes that happen.
It's like mining for that gold so we know how to.
You bet.
Reprocess it.
Yes.
I think the ability to have good data and have the right guardrails around it is also going to be key to what we do with AI. We can see the outputs, right?
Oh, right on. Yes.
Yes, absolutely.
I think we're starting to hear a little bit more about what Cisco is doing uniquely in the AI space in terms of building their own models. Not going to ask you to comment on that because I was like, I'm just.
Oh, no.
Like.
Do you have any thoughts?
Oh, like AI Canvas. Oh my gosh, that's ripped out. That is like crazy. And I just have a strong hunt. Splunk's going to play a role power in that.
That's perfect. Jim, I think you forget, this is your first Cisco Live, right?
It is.
The good impression.
This is awesome.
Yeah.
Oh, yeah. I could just sit here and listen to the tune.
It's like you had like a little bit of a shot. This is a lot. This is a big content.
Yeah. There's a lot of human beings. Yes.
That's awesome. Thank you for everything that you do. Guys, we'll go back to you in the studio.
Robb, thank you so much. Our thanks to Jim as well. Good to get that Splunk story. As Jim just said, the Mac daddy on the data front. I actually really love that. Speaking of that, I'm going to bring in the Mac mama who is sitting not too far away from me because Lauren, you have a lot of direct Splunk experience. You came to all of this on the Splunk side. What was that experience like in terms of the move to combine the data power, the power of the aggregation of the data lake and the application of that data on the Cisco side? Give us your insight.
I honestly think that it was a match made in heaven. I know that sounds like maybe that's too much, but it's true, right? Because if you think about it, I love what Jim said about Splunk being the Mac daddy. I feel like nobody knows data like Splunk. And then nobody knows networks like Cisco. And then nobody can power this thing with AI like Splunk and Cisco together. For me, it's really been amazing seeing the technology seamlessly integrate. The people, just being able to share like the shared vision, working as a Cisco employee for a number of years and as a Splunk employee, I will say that both companies have the same values: people-centric, customer-centric, and problem-solving-centric. Again, match made in heaven. I love seeing us be better together. We may overuse that, but we're doing it because it's true.
We truly are better together when we come as one. You are seeing it through all the discussions. If you go out in the world of solutions, you will see the power of the one platform. You are not hearing technology by technology. You are seeing outcomes. You are seeing how we are driving business outcomes.
It's a great, great point, Lauren. I appreciate it. I want to go back to something Chuck said yesterday in the opening keynote. I'm going to quote him directly because he's going to say it better than I ever could. "Between technology transitions and the geopolitical landscape, our lives are so much more complicated. And the demand to build with security, integrity, and consideration for humanity is larger than it's ever been before," right? That is the idea of putting these great minds together. That security track is so vital, vital. Also bringing the best minds together. That's what we're going to go out to Michelle right now, who's got one of our NetVet and CCIE folks with her. Hey there, Michelle.
That's all right.
Michelle, you got me?
Hello. I can hear you now, Steve. Thanks for bringing it over. I'm with my new friend, Keith. Keith, how many Cisco Lives have you been to, and where are you coming from?
I have actually been to seven Cisco or Cisco Lives, even back when it was Networkers back in the day.
An OG?
Yep. I am from Cedar Rapids, Iowa. Yep.
Welcome. One question for you. What was most impactful about all of the stuff that you heard yesterday, and what are you looking forward to today?
All right. To me, the dashboards, being able to troubleshoot things in a really quick manner, that is an awesome thing to have. I am just looking forward to how AI is going to take us to the next level.
That's exactly what we're hoping to hear as well today. AI is a big part of today, this keynote, as well as the rest of these sessions. Stay tuned. Tune in. We've got you. All right. Now I'm walking over to Z.
Thanks, Michelle. I tell you, I am here with Josh Dodd. He is from sunny San Diego, right here from the area. I tell you guys, Josh is expensive because he came from the Splunk acquisition. This man is worth, you know, he's worth more than I can even pay for. Josh, tell me, what brought you to Cisco Live this year?
Thank you. First of all, I hope we're worth it. This is my first experience at Cisco Live, and it's been electric. It's been wonderful to be around the Cisconians, seeing the customers. I loved the keynote yesterday. I think I especially appreciated Jeetu talking about different services, product launch announcements, and of course, the integration of Splunk and how Splunk's going to play a key role in serving our customers.
Yeah, absolutely. Josh, I appreciate your time and look forward to much, much more. I tell you, keep watching, lean in, because there's more ahead of you.
More to come.
Yeah.
Thank you.
All right, Z, I'm going to take it back over here in the studio. I appreciate it. Our thanks to Josh as well. Welcome in from the Splunk side. It's good to have you as part of the family. Lauren, we're about, we're going to call it two minutes from kicking off the keynote. We are so excited about that. A moment ago, Michelle had a chance to talk with one of our NetVets, one of our CCIEs. Education is such an important part of this event, right? Not only do we come together to learn about the Cisco products, technology services, AI, security, networking, but validation is an important part. Getting Cisco certifications, they're free here at the show, right, Lauren? How cool is that?
It's super cool. There are the labs, our hands-on. We have DevNet. The learning doesn't stop. If there was ever a place where you can learn as much as you need to and want to, this is the place. What makes it the most, like what makes it the best is that it's fun. It's fun to learn because everybody around you is also excited to learn. You don't feel like, I don't know, like a stamp, like, "Oh my God, am I the only one?" You don't feel that way. I think I said it earlier, but you're never alone here at Cisco Live.
Oh, that is such a good point. Good point. Lauren, that has brought us right up to the vital point. I'm going to go ahead and make the introduction here for the keynote. We are so excited to head on in. This is the day two keynote here on Wednesday morning from San Diego. Let's set this up. Innovation in action. Chuck Robbins, Chair and Chief Executive Officer. Carrie Palin, Senior VP and Chief Marketing Officer. Liz Centoni, Executive VP and Chief Customer Experience Officer. Oliver Tuszik, who I get to talk to later, Executive VP, Global Sales and Chief Sales Officer. Huge minds, huge talents, great leaders behind Cisco's drive to harness the power of AI. We are going to see the ways that we can drive growth, build a better world together.
We will learn about all of the new innovations, the new announcements, the new releases that we've created here in the AI era to really hone in on the Customer Experience. How Cisco helps you, your organization navigate through it all. It is so hard to navigate right now. Things are changing faster than ever. We can't keep up with the pace of change. How far out can we really plan before we have to rewrite the playbook over and over and over again? You don't have to do this alone. Cisco is your partner in every single way. We create the tools, the technologies, the services, the solutions. The people who are here are visiting all of those down at the Cisco Showcase where we'll tour a little later following the keynote. It is a very exciting day. Special guests coming in as well.
From LangChain, University of California, San Diego, from Ford Motor Company. We're going to be heading out to the keynote in just a moment. As a reminder, as the keynote rolls by and you hear these great insights, these great thoughts, ping us with your responses. What did you love? What did you get excited about? Put them up there on social media. We want to hear what you have to say. What is getting you excited? Where are your thoughts going? Let's head out to the keynote. Away we go.
Please welcome back Senior Vice President and Chief Marketing Officer, Carrie Palin.
Good morning and welcome to the day two keynote. Yesterday, we heard so much about that massive innovation payload that Jeetu and his team dropped on us right here on this stage. I spent the entire day thinking about all the notes I took, all the excitement I had with everything that we're going to be experiencing in the next just a few months. Today, we are actually going to hear from customers and partners directly as to how that Cisco Innovation is driving really important outcomes for each of them. First to the stage, we'll have Liz Centoni, Cisco's Chief Customer Experience Officer. With her, she will have very special guest, CEO and Co-Founder of LangChain, Harrison Chase. By the way,
I just said Harrison Chase, and I thought forever I was going to say Harrison Ford. I got it right.
I'm so happy about that. Harrison Chase. And they're going to be discussing the future of Agentic CX. Next up will be Oliver Tuszik. He is our EVP of Global Sales and Chief Sales Officer, and new to role. We're so excited to have him in the role. He is going to be bringing two really cool customers up to stage with him. One is Ford, and he'll have Patrick Milligan, the Chief Information Security Officer from Ford. The next is the University of California, San Diego, with Dr. Vince Kellen, CIO. I think the most important thing about today is hearing everything about how Cisco's technology is actually driving important outcomes for our customers through their stories, through their experiences, because that's why we do what we do. We do this for you. I'm excited to hear from them.
I hope you all enjoy the keynote. I hope you have an amazing day afterwards and a wonderfully fun night with The Killers tonight on stage. Mr. Bright Side, anyone? Right. Okay. I'm excited about it too. With that, we're going to bring our Chief Customer Experience Officer, Liz Centoni, to the stage.
Good morning, everybody. Wow, good. A crowd that's awake already. I love that. Yesterday, you heard from Jeetu about some incredible innovations. I want to make clear, our job in CX is to bring that technology to life in your environment. I'm going to spend some time today talking about the investments that we're making to better serve you in the AI era. Now, everything that you're going to hear from me today did not exist a year ago. It's brand new.
This is incredible in terms of the speed at which the team has delivered. I'm also excited that you get to hear directly, and we get to learn from Harrison Chase, the CEO of LangChain. The LangChain team has really been instrumental in helping us move from just experimenting with agents to actually building useful agents at scale. Let me start with what you've been telling us for years. It should be very familiar. You want resiliency, simplicity, and time to value. Now, in the era of AI, your expectations have just accelerated even further. You've been telling us that, "Hey, I want you to move rapidly from break fix to give me proactive and predictive experiences. I want hyper-personalization from deployment to adoption to support across every touchpoint that we have with you.
I need you to speed up time to value, but with human oversight and trust. When I think about the traditional approach to Customer Experience, it's just not built for this level of complexity or pace. What's needed is a new kind of CX. Over the years, we've thrown automation at it. We've layered AI on top, and we've made progress. Actually, to be fair, really good progress. It still needs humans to stitch everything together. That's the gap. That's where Agentic Systems come in. The exciting part about Agentic Systems is it gives us the ability to solve problems that we all have been just circling around for many, many years. problems, which I say boring problems. My team hates it when I do that. It's problems that we all know that have been a hassle for so many years.
We want to change the nature of how we interact with you, where we go to intelligent anticipation versus just pure reaction, where we can do radical personalization, but do it at scale, and where we can amplify our team's expertise, the teams that support you by removing the friction that they face in doing their jobs every day. Now, using Agentic Systems, over time, these systems understand operational environments deeper than any one person can because it is continuously analyzing historical data, real-time telemetry, interaction patterns, and feedback loops across the full stack. What it helps us do is it helps us understand your environment and your teams better so that we can personalize, we can anticipate, and recommend actions. Our goal in this is to make every customer feel like our only customer.
Now, to deliver this amazing experience, we've got to get rid of the problems that actually stand in the way. Like problems like where 50% of our network professionals' time is spent on manual tasks every single week. 80% of security breaches are actually caused by human error. Now, addressing these pain points requires more than just incremental fixes. The AI era isn't just changing what we build. It actually is opening up entirely new ways in how we deploy, adopt, and operate technology. The way we're approaching it is we're fundamentally changing how we deliver resiliency, simplicity, and time to value. When I think about, okay, I was thinking about this and saying, how do I go deep into each one of these and still keep it pretty crisp? My mental model is things in three.
I had to actually think through which are the three favorites that I pick under each one: resiliency, simplicity, and time to value. There's a lot more that we're doing that when you go to the CX booth on the show floor that you'll get to see. Let's talk about how we keep your environments highly resilient and reliable. Your configs can feel like the new tech debt. And why do I say that? We get about 1.67 million support cases every year. 25% of that are related to config issues. We can envision a future where we go from config chaos to config confidence. Like with services as code, we're integrating pre-testing with the deployment pipeline so that we can catch config issues and errors upfront. Now, make no mistake, this is not just about finding defects.
Here, we're proactively validating configurations against established best practices and your specific operational requirements. Our goal is to be able to give you high confidence and proactively mitigate risk. We do this with ACI today. It's in deployment with over 100 customers. We're rolling this out across the entire Cisco portfolio. Next, when I think about network drift, I think about it as submarine warfare. It strikes below the surface, and by the time you realize it, it's done some sizable damage already. We all know that networks drift over time from their original design intent. Our goal is to fix it before it falters. We're doing this with a continuous digital evaluation of the network's current state versus its intended state. We're then using AI-driven recommendation to remediate and align with best practices.
The third thing that we're doing here in the area of resiliency is around building real-time health checks to include things like cross-domain dependencies, where we can do the proactive monitoring and alert you or provide you with early insights into potential issues in your environment and provide recommended actions as well. Let's do a little bit of a sneak peek in what services as code is. Please roll the video.
40% of network outages, 9% of annual revenue lost, all because of configuration mistakes that never should have happened. You are going to see how we are turning configuration chaos to configuration confidence and the first steps of that journey with services as code. Imagine you are remapping bridge domains within an ACI fabric. While updating the VLANs to voice, you make a mistake in the configuration, forgetting to change VLAN 103 to VLAN 203. Now, with services as code, you get a best-in-breed GitOps with integrated tests. When you commit the changes, the system detects a mismatch between configuration and intent and fails the commit. Now, aware of the issue, you're able to go back to the editor, fix the configuration, recommit the change, and merge the request, triggering the DevOps pipeline and ensuring successful deployment. One of our customers who adopted this summed it up perfectly.
Security, resiliency, consistency, you delivered all three.
Now, I get excited after seeing that. There's a full demo that's available in our AI-ready data center booth on the show floor. Next, I want to talk about how we're helping simplify the experience. How do we make it consistent and easy-to-use automation across the lifecycle? You've given us that feedback many, many times that we need to simplify the experience with fewer interfaces and more validated integrations. We're providing you with a single and intuitive experience. You'll get a hyper-personalized view into your entire Cisco environment, where you can see your install base, the adoption progress, risk analysis, and you get integrated troubleshooting as part of that as well. You can actually generate dashboards programmatically. You can see your software renewal timelines. You can see vulnerabilities, and you can see misconfigurations as well.
The other thing that we're doing in the area of simplicity is dynamic change assistance. Here, we're focused on helping you evaluate the potential impact of changes in your environment, changes like performance shifts or field notices, and provide guidance to be able to minimize risk. The third area that we're doing is we're digitizing this massive intellectual capital that we have with tech cases and professional services best practices so we can proactively alert you to opportunities around optimizing your environment and reduce the likelihood of outages. As you'll see across all of these, these are pain points you deal with every day that we help you with as you reach out to us on support or even on deploy and adoption. It's not a lot of rocket science here.
I would love for us to show you a sneak peek into what that unified interface looks like. Please roll the video.
They say you can't have hyper-personalization and simplicity together. Pick one. We say, why choose? What you are about to see is the first major step towards a unified experience that gives you insights across your environment and learns your operations inside and out. It's the beginning of building something that eventually knows you better than you do. Taking a look at your Cisco assets, you see what's in use and what's on its way out. The last date of support grabs your attention. A curated list and timeline shows which devices are affected and when, helping you plan and act confidently. The risk explorer enables you to proactively manage risk unique to your environment. From the overview, you spot a vulnerability tied to a new security hardening guide. This lengthy technical document has been transformed into a clear set of compliance checks tailored and run against each device.
Selecting a device, you see the unique diagnostic information. The system generates a script to remediate the issue, and you can execute it with ease. The system reruns the compliance check to confirm the fix, improving your security posture. In the support experience, you see your cases and history. Additionally, you can access the unified AI assistant for troubleshooting, which builds on the Cisco Support Assistant available today. With personalized context, you can ask questions about devices in your inventory and receive tailored responses. You can even ask for a recommendation to address a security advisory, allowing you to quickly resolve issues before they become a problem. Our mission is simple: achieve the depth and breadth of understanding where every customer here feels like they are our only customer.
That's the exciting part for us. We want everyone in this audience to feel like our only customer because as Agentic Systems come in, we want to know you better than you know yourself and keep your environments highly resilient, reliable, and simple as well. The third one I'm going to talk about is how do we help you maximize value from your Cisco investments? There's a gap. We all know there's a gap between purchased and value realized. We want to be able to close that gap. What we've started doing is we're digitizing your intent, your goals, your KPIs from the start so that we can prioritize features that actually matter and align to your intent, not just a set of features that we give our customer success teams.
This way, we can also adapt in real time to align to your evolving needs because we know that's not static across the full lifecycle. We just released an adoption agent that provides a personalized adoption journey, aligning onboarding to your intent directly. Next, we've also created this notion of a personalized automation framework. It creates a validated design approach based on our understanding of your environment, the design, the real-time telemetry. What does this do? Say you want to add a branch to your new environment or to your environment, we ensure that the SD-WAN and the unified branch operations configs are tested and validated. Lastly, when we think about mean time to repair, maybe we have a different view of what that means. I look at MTTR as a measure of how much manual work that we're still doing.
Over 60% of our attack cases, and I believe the number is much higher now as well, are resolved leveraging AI and Automation and helping us shrink the time down to issues resolution. Our attack assistant, which has more than 250,000 unique users, works alongside our support engineers to offer remediation suggestions. Think about what's the next best action. This is based on extensive knowledge base that we've digitized and best practices from resolving similar issues. This is a big shift. It's a major shift from manual workflows to proactive AI-assisted automation. We're embedding this AI assistant support into our products. You'll see this down whether it's digital resilience, whether it's in AI-ready data centers. We're embedding this into our products to streamline operations even further. We're not just accelerating time to value. We're transforming how quickly you can turn intent into real impact.
In summary, I would say is we're building a future-forward CX powered by Agentic AI across the entire lifecycle, across your entire lifecycle. The goal is to give you foresight, faster adoption, and greater value every step of the way. Agentic CX is not just about better service. It will certainly provide you that. It's an opportunity to grow with you, to learn from you, and to adapt with your needs as your needs evolve and the world continues to change. Now, building this future, we're doing this with a few key partners. One of them is LangChain. Please help me welcome the CEO and Co-Founder of LangChain, Harrison Chase. I'm going to start with a few words on Harrison because if I ask him to introduce himself, he'll probably say a one-liner. He's way too humble for that.
He just did not jump on the agent bandwagon today. He built the framework, the tools, and the community that made it possible for the rest of us. I am sure many of you in the audience have benefited from him and his team's work and do not even know it today. LangChain, for those of you who do not know, solves the biggest headache in building useful agents at scale, making them reliable, and connecting them to real systems. Millions have the same problems, and you had, what, 70 million downloads last month? Harrison has been hands-on in the evolution from simple chatbots to ambient agents that can orchestrate complex workflows and integrate with the enterprise tools. Harrison, great to have you. Thank you for joining us.
Thanks for having me here. This is my first Cisco Live. It's awesome. Carrie, I apologize that I'm not Harrison Ford. He is the superior Harrison.
I think you'll always remember that one. I will every time we'll watch an Indiana Jones movie. I want to start with something that really resonates with why we chose to work together. You could have built LangChain as a closed platform, probably made a lot more money. Instead, you chose to open-source the foundation framework. What drove that decision?
I think there's a few reasons. I think one of the core things that we believe is it's really early on in this journey of GenAI and LLMs and agents. We think that by working with an open-source community, we can just learn a lot more and develop the framework and the ecosystem a lot faster than if it was closed source. Optimizing for this speed and learning was a huge part of that. The other part is just where LangChain sits in the ecosystem. We're the glue that holds a lot of these components together and orchestrates them. I think something like that in today's age just has to be open-source for the developers who want to use it.
Love that. Your open approach aligns perfectly with our agent future for CX. We're thinking about agents everywhere across our many workflows, and we have a lot of those. I also want our teams to have freedom to pick the models, to architect the intelligence. Can you talk some more about what this means? The audience is looking at the slide where I did lift some stuff from your conference, by the way, that's on the slide, so kind of plagiarized a little bit. Talk some more about it. Also help us understand how should we think about monitoring and troubleshooting? Because when I think about a distributed agent system and I think about, hey, it's the Customer Experience on the line.
Yeah, this is one of the big things that we've been focused on because it's pretty easy to build a demo of something that looks great on Twitter, but getting it to be reliable enough to put in front of thousands or millions of customers is a whole different thing. When thinking about how to make GenAI agents actually work, there's a few different kind of processes that we like to think about. First is kind of like just the debugging workflow. When you're working with these agents, when they go off the rails, they go off the rails because what goes into the LLM at the end of the day just doesn't have enough context or information, or it's not kind of appropriately specified.
Being able to see all the steps that happen in the agent, all of the different retrieval things it does, the sources it's gathering before it calls the LLM, that's super important for this debugging workflow. Being able to see exactly what goes in and exactly what comes out is super important. Crucially, this isn't just developers that are trying to debug these agents. This is oftentimes subject matter experts or product managers because those are the people who know the agent the best. After you debug a single thing, you then want to go on to kind of testing it more at scale.
Building up tens or hundreds or thousands of different test cases that it might see in the wild that are representative of what people might ask it to do, and then run against these to benchmark, make sure you're not regressing, things like that. Finally, when you deploy it and you're getting thousands, millions of queries, you can't look at every single one of them by hand. It becomes time to try to look in aggregate for statistics or trends. What tools is it calling? What paths is it going down? These types of production insights are a third type of debugging and insights that you want to be generating.
Those orchestration tracing capabilities become a lot more important. Not that they weren't before, but this is a different way in terms of how we're approaching that. Now talk to us a little bit about, we talk about the fact that the future is really mostly about ambient agents. What does that mean? Why should we be excited about ambient agents?
This is one of the things that we're most excited about. If we think about AI today, we mostly think of chatbots that we go to and we message them, and then we see some text streaming back at us, and then we message them again, and there's this back and forth. That's great. That's useful. I use ChatGPT or Claude every day. I think there's a few reasons why we're excited about ambient agents. Ambient agents we define as agents that are triggered by events, run in the background, but they're not completely autonomous. I want to break down those things. If they're triggered by events, that means that I can scale myself as a professional a lot more because I'm no longer kind of kicking off an agent. There's not a one-to-one kind of interaction. It's now triggered by events.
There could be thousands, hundreds, thousands, millions, whatever of agents running in the background at the same time. I can scale myself and scale the effect that I have a lot more. The key part is running in the background as well. If I have chat open, maybe I can have two windows open side by side, but again, I'm kind of rate-limited by just the interactions that I can kick off and be looking at. If they're running in the background, it's a little bit different. That also lets agents do a lot more interesting things because in the background, they can take longer to do things. Crucially, I think ambient does not mean autonomous. We very much believe in human-in-the-loop interaction patterns.
If you have an agent that's running, there's absolutely going to be cases where you're going to want to give it feedback, where you're going to have to answer questions, approve actions that it's going to want to take. I want to also emphasize that ambient very much, I don't think, means autonomous.
I'm glad you said that because we look at it as augmentation, not taking over roles, maybe task at some point in time. Now you see through LangChain millions of workflows. So what's the real scoop between real and marketing? How much of today's agents are just sophisticated chatbots?
I mean, it's still early on, but I think we're getting there. If you look at, coding is a vertical that I think leads the way in a lot of agent and AI applications. If you just look at what's happened in that industry over the past year or so, I think it speaks for where we're going. We started with kind of just autocomplete of individual lines. Then we moved into kind of chat experiences in the editor. In the past three weeks, there's been five or six different long-running async background agents that have been released from OpenAI, Anthropic, Cursor. I think this speaks to the future of where we're headed. We're headed more and more towards these longer-running things that run in the background. You can kick off many in parallel.
Yes, I think it's early, and most things are chatbots, but I think that's absolutely where we're headed.
Something you said just stuck with me in your prior answer. It means that we need to flip our mindset as well. We can't just think about the traditional ways of build it, ship it, and then maintain it. It's almost like our mental model needs to shift to build it, learn from it, and continuously improve it. That goes for not just developers, but users as well. All of this only matters if we execute. Maybe help us understand a little bit what should Cisco customers and the audience expect from what we're doing. I know you touched on this a little bit, but maybe package it for us.
Absolutely. I think you touched on a few key points as well. One is just model optionality. There are a lot of different models out there, and they're good at different things, and you're going to want to use them in different scenarios, whether it's on-prem, whether it's really small ones that run on device, whether it's larger ones that run in the cloud. Through LangChain and through our collaboration, Cisco customers can expect that whatever agents come out there will be powered by the appropriate model at the appropriate time. We're also working on these types of longer-running background proactive ambient agents. A lot of this is powered by LangGraph, our framework for building these types of agents that emphasizes the long-running and the human-in-the-loop interaction patterns. Then we talked about debugging, of course.
This is where one of our tools, LangSmith, comes into play. It has that whole debugging, testing, and evaluating, and then monitoring thing. Crucially, I also want to say that one of the things that we're really excited to work with you all on is making sure that all of these things and all of the agents that we build can run wherever Cisco customers want them. That means on-prem, that means on-device, that means on the cloud. You guys obviously have a very storied tradition of on-prem and bringing things there. We're excited to push boundaries there as well.
Wonderful. I'm glad you emphasized the on-prem aspect of it as well. I'd say in closing, we're past the point of asking if agents will transform CX. The only question is how fast and who leads. The infrastructure is here. The patterns are proven. The teams that architect the intelligence today will define what Customer Experience looks like tomorrow. Thank you for showing us what's possible when you stop talking and actually start building. A round of applause for Harrison Chase, please. Thank you.
Thank you, Liz.
I feel every time I sit down with him, I learn something new. He ends up making us a little bit more smarter each time. I want to thank you all because everything that you saw here is your influence, your feedback. Thank you for helping us as we jointly walk this future together on what an Agentic-led Customer Experience looks like. When I think about Customer Experience, is the customer at the center of everything that we do? I am excited that we now get to hear from Oliver Tuszik and two of our most favorite customers. Please help me welcome Oliver Tuszik.
Thanks a lot. Thanks a lot. So good to be back in the U.S. So good to be back here at Cisco Live. Let me come back on what Les just presented to you. I hope you saw we're not talking about the vision of AI. We're talking about embracing it, utilizing today. This is a real-world use case delivering benefits, improved quality, faster solutions, and productivity gains right now. This is an amazing use case within Cisco. If you want to hear more about this one, want to learn from our experience, or want to get help on how you can embrace AI, please reach out to Liz and the CX teams, to the account teams, or any of our amazing partners out there. I hope you can see that Cisco as a company is becoming a 100% AI company. I know that's a big statement.
Do not get me wrong. I'm not sure if Kevin is here. We're not becoming a competitor for OpenAI. Everything we do in every part of the business, we embrace AI. The way we design our products, the way we embed AI into our products, the type of products we develop to allow you to embrace the power of AI in a safe and secure way, how we deliver and build our services, how we run our operations, how we manage the entire system, we embrace in every piece of this AI. That's how we want to lead in this area. Now, I got a pretty tough job at the end of the session to summarize all the announcements, the amazing innovation, the new features, everything you saw over the last days. That's quite a lot. And G to spend a lot of time on this one.
I don't believe I will be able to do this just in five minutes. What I want to do, I want to explain what is behind it, what is driving us, how we are building this portfolio, and I mean the entire portfolio, starting from the core networking, security, collaboration, observability, data center, AI, and of course, not to forget the amazing piece of Splunk that we embedded in everything. What is behind this? What is the driver? Jeetu is always talking about creating products that customers love to buy, that are 10x better, focusing on quality. At the end, what is driving us is simplification. We are driving simplification, building more integrated products, building platforms because you want to reduce your complexity. You want to increase your productivity. We help you to create digital resilience across the entire portfolio.
Of course, we help you to drive the adoption of AI. There is one more thing. One more thing that is overruling everything. It is a fundamental principle how we design our product and services. Everything at the end starts with you. Everything we build, everything we design is built based on your feedback. No matter if you are a CCIE running operations, running a SOC, your needs define our roadmap. Your challenges are what we are focusing on. Your vision, your mission, what you are trying to achieve is driving us. Whatever we do starts with you. If I want to end this session, I think the best way would be to bring all the customers up on stage and let them talk about, did we deliver? I do not have the exact number, but looking at the stage, I think it would be pretty crowded.
Instead of asking all of you to come up on stage, I will pick two customers, two amazing customers that we have, customers that have been working with us for a long time. The first one is a local customer, a customer that is running a pretty complex, challenging environment, a $9 billion business. I had a couple of discussions with him recently, and he was telling me how he's driving the AI motion. He made a statement that I want to share with all of you. It's a pretty big statement because he said we should aim to respond to the prediction, not to the incident. He's moving the automation to the next level. He really drives this with a power I've never seen before. It's a great honor for me to welcome on stage the CIO of the University of California, San Diego. Please welcome Dr. Vince Kellen.
I want to than k you for the glowing comments, but when my wife hears that, she says, "That's great, Vince. Take out the garbage."
Before we get to this pretty bold statement and your view on AI, maybe tell us a bit more about your role and UCSD .
Yeah, I'm the CIO. Think of me as kind of corporate CIO. We have a CIO in the hospital. We work hand- in- hand on all of this, especially the network and security layer. I'm responsible for the larger hospitality mission, the research mission, and the education mission, and of course, the administration for everything. It is a lot of technology. We have 15,000 access points, 5,000 switches, firewalls, galore, use of Splunk. It is a complex environment. It is kind of like a city, actually. We have at least 100,000-150,000 people going through our campus each day.
You've been talking about a city. Now let's talk about the citizens. You were telling you have some of the most demanding end users. I heard stories about even active protests if something on the IT side is not working.
No, that is correct. Some time ago, I talked to our CISO, and I said, "We have to protect ourselves from the outside world." She's like, "We got to protect ourselves from our dorm students." They're a demanding audience. You take some Maslow's hierarchy. Wi-Fi is underneath, and underneath that is battery. When those are gone, they complain. It is a very tough environment. We also have rogue stuff coming in all the time on the network. It is a complex, dynamic environment.
Talk about complexity. I think higher education in these days is essential for all of us, but it's also going through some changes and challenges. Talk a bit more about how you look at this and how you look at the future of higher education.
Higher ed, as we all know, has had some strong pressures over the last decade: the demographic cliff of reduction of high school students, the need to get to professional education, and of course, all the funding challenges that have been growing. All roads are leading to high levels of automation everywhere where we can, but at the same time, enhancing the student experience and the quality of the education, a very tough task.
We've been talking yesterday in many of the discussions about the usage of AI agents within the companies. There is a lot of companies that are afraid of losing control. Do you have a set of regulations or controls, or you limit the usage of AI agents within the?
I won't say we limit, but we certainly think deeply about that regulation. I think of regulation at many levels from what I call societal norms or ethics and bias to policy and to law, as well as within agent frameworks. How do we regulate the agents, the tools? How do we regulate the tools? The tools that call the other systems, how are they regulated? You actually have a rather robust, interesting regulation environment. I don't look at it as an obstacle. I actually look at it as a very interesting challenge for which this new agentic framework is going to give us a lot of tools to address.
You made a comment yesterday that it takes about nine months to write the regulations. I don't know how long it takes to implement, but once it's done, the technology has changed.
Absolutely. We're aware of that. In fact, some of the faculty we're working with right now are trying to develop policies on research. They recognize this is a very dynamic area. You still have to go through it as an organization. The organization has to come to grips with what AI is. At the same time, as CIO, I can't let that stop progress on making the environment safe and performant.
Let's talk about your kind of setup, installed base. You have a lot of Cisco gear. Looking at all the announcements, how we bring networking, security closer together, and embrace AI in everything we do, how are you looking at this? Does this help you? Is this what you expect from Cisco?
Oh, this is exactly what I expect. Not to be sort of unkind to Cisco, but I summarize the product strategy in one sentence: Skynet. Not quite because human in the loop, and that's important. We absolutely need this level of automation and automatic, a sort of self-healing, curing environment, always with a human in the loop. We need to do it now in order to address, one, the demands on us, but also the funding challenges ahead.
Let me double down on this one because, as I told you, I really love the quote. Tell me a bit more. What is behind this idea and how you're trying to get to the point that you've just been able to react before something happens?
Absolutely. We all know once you're in reactive mode, it's taking 5x, 10x more time than if you're predicting. If you have the blast radius and you nip it, but even before it becomes a blast radius, you save oodles of amount of time. Just the Canvas product that you've come out with is really interesting because it's going to allow very rapid discovery observability. It is also setting the seed now for great proactivity and, in fact, perhaps adding humans as agents and a digital twin of that human within that agentic framework so the knowledge and the experts that we have can get into the AI, relieve them of some of the burden of constantly replaying their oral history and knowledge back to us. We see this as kind of a symbiotic system that, in time, will really improve its efficiency.
We'll be able to get real good scale with very little mass.
I hope you can see that DJ and G2's team working exactly in this direction. Coming back to the security piece, and I think being such a large university, you're the perfect target. I think everybody is targeting you right now. How do you handle it? We're still keeping up a kind of open-minded environment because you have some of the smartest brains in the world coming to the university. You do not want to limit them. How do you balance this challenge?
That's a difficult balance, obviously. The way to do that is through great user experience. All researchers want to be safe. Everybody wants to be safe. What they don't want is have it constantly wagging them in the face like a tail on a dog. We work hard with all the Cisco tools we have and other tools in order to make that experience as pleasing and out of the way for the researcher. It also requires this much more defense in depth thinking. When I talk about regulation, it's now at all levels. It's at the device level with the agents on the device. It's at the enforcement level of the network. It's at the inspection level of the packets in the network. It's now into the agentic framework and the AI.
This defense in depth really has to be exploited because we can't really do simplistic, monolithic controls over the whole shooting match, especially with the students.
Now, looking at security in the days of AI, do you see a radical change on the attack side? Is something changing? How do you predict will it develop over the next six to 12 months?
On the attack side, in the last five years, we've seen a significant change. We are a big target at UC San Diego by virtue of our research. We got an oceanography institute. I've discovered when you put sonar in the water, you just detect more than fish in the water. Other countries want to know about that. We are a big attack vector for other parts of the world. This really caused us to have to really up our game in security. Think of it as a pyramid. You're going to have your petty thieves. That's the attacks that we kind of saw five years ago. In the last year or so, we've seen very exquisite attacks coming out of foreign actors that are pretty well financed.
What you really have to do is you've got to cover all that petty thieves stuff and get that out of the way. You've got to get your two-factor, and you've got to get your zero-trust frameworks going. You've got to get this autonomy going within the system so that now you can focus on the really exquisite attacks, which tend to be long-term. There's something in your environment that was a stray, and that's really been done maybe a year or two ago, and now it's coming back to bite you. That's where we're going to have to focus our effort.
Now, coming back to your journey with Cisco and how you utilize Cisco, maybe you can share what makes Cisco special for you or what do you like in this partnership?
Yeah, it's become apparent, certainly to me and others, that this notion of end-to-end, top-to-bottom observability and integration, and now AI as an integrating force is really critical. The time and the latency in that becomes very critical. The engineering that companies like Cisco are doing is fantastic. We'd like to see more of that in order to make all of this work. The other piece is support becomes a big deal. We were an early adopter to the ICE framework for wireless, and we really tested the bejesus out of it for Cisco several years ago. They were behind our back every way, even when we had hiccups along the way. That support element really does loom large for many organizations, certainly for us.
Talking about AI, and I think you saw the AI Canvas. I got it right. It's not going fast and far enough for you, but you told us this is exactly what you're looking for. Even you want to have it more or less fully automated.
Yes. I think the Canvas framework is great. My suggestions have been we have all this knowledge in human beings around the network policy, et cetera, that we have to get out of the minds in sort of a drip irrigation fashion. We need to bring it into the AI. We need the AI to improve its performance over time based on this human-technical symbiosis, which has to occur. That is going to let you sort of ramp up and get the AI to learn without having to resort to all of low-rank training or new model training, et cetera. There is enough data and content that is prioritized in all of this that that can bootstrap you up into higher levels of performance. I think that is where Cisco is absolutely going.
I think this digital twin concept, not just digital twins within the switches, but human-digital twin, get my engineers, get a digital twin of them, call the engineer when there's an incident, engineer confesses knowledge, that gets put into the AI, and that person doesn't have to be tortured yet again for another incident. I think that whole process can start in this coming future.
That's a good one. I'm not sure if Jeetu and the teams are working already on a human-digital twin to do the daily job.
They have.
Maybe next Cisco Live.
Canvas have a skill in there that says, "Hey, Bob and Mary are experts in this area," and it actually routes to them. They confess in verbal form. It's scraped, converted to text, augmented, thrown into the hopper. And now they don't get called the next time.
Perfect. There was one topic that was also discussed more often over the last day. This is the question, do we focus 100% on cloud, or is there a new world, especially with AI, some data privacy issues where you should shift a bit more to a hybrid or on-prem version? I know you have special requirements, special logic, but maybe it's interesting for the rest to share.
Yes. I have long been a proponent of cloud. What is happening with AI and the cost structures and the demand, and more importantly, what I call full-stack sovereignty, right from network, data center, data, AI, wireless, and satellite, you are seeing this full-stack sovereignty, which now opens up a really interesting distributed computing environment between rem and Cloud. You throw agents on there and agent-to-agent communication through those standards. I think we are entering a new era of massively decentralized AI computing from the phone, the laptop, the workstation, the switch, the routers, every part of the network. That means we are definitely in a hybrid world for which infrastructure as code has to be the way to push all of this seamlessly in any direction, whether it is Cloud or on-premise.
Now, moving on and talking a bit about budgets, because budgets and how you pay for all the needed investments is a big challenge. Based at least on my understanding, I have not seen any announcement of additional funding for the higher education. How do you balance this challenge that you need to drive short-term investments, especially to drive automation with the challenge that there might be no extra dollars?
Absolutely right. As a CIO, I'm thinking on the investment lens of this. One way we think about investment is we're going to do something. It's going to stretch out three, five years. We pile in the investment, and we get our payback in that time frame. We are shortening that time horizon to less than a year, hopefully months, which also causes incrementality to increase. This notion of continual deployment arises, smaller projects. What's nice about AI in our analysis of all our AI work, not just on the network, but in the application layer, use cases are pareto-distributed, meaning there's a few use cases that drive enormous business value. There's a lot of use cases that have little value. Harvesting that value, I'll give you a metaphor. On my property, I've got a grapefruit tree and an olive tree.
I can go to my grapefruit tree with a basket in 10 minutes, pick grapefruits, fill it up, can bring it back home, cut them up, and in another 10 minutes, I am drinking grapefruit juice. When I pick my olives, I'm out there for six hours. I barely fill half the basket. I got to bring it back, clean the olives. I got to wash them. I got to brine them for a week. Got to rebrine them for a month, repeat that process another three months, pack them. Four months later, I can have an olive. It's not quite that long with AI. The moral of the story is getting value out of AI is more like picking olives and grapefruits. There is a long tail here, and there's a lot in that long tail, but a lot of it is mundane and simple.
That is where we're starting. That lends itself nicely to shorter time horizons, quicker paybacks, quicker wins. Still want to tackle the fat tail of the AI use cases. Boy, this is beautiful for the long tail.
Perfect. Maybe as we're running over time, last question. Any advice based on your experience on how to adapt and how to move fast in these times?
First, do not listen too much to what you read on either side of the equation. Use your own judgment and your own mind. Agentic is here. There is no question. AI is here. There is no question. Start the work, begin the experimentation. Start with the mundane. There is so much in the mundane. You can consider incidents the mundane. My security team do not like me saying that, but I view it as a CIO as the mundane. Let us automate the heck out of that. Let us get on to more profound things from here.
Perfect. The summary starts today. It's here.
Yes.
Big applause, Dr. Vince Kellen.
Thank you, sir.
Thanks a lot.
That was amazing.
It's great to see that the higher education is that far ahead on this one because that's what we need all over the world. My next guest is a long-term Cisco customer. I think he's the longest-serving member of Cisco's Global Customer Advisory Board, and mainly because he's very clear, very outspoken. Together with him and the amazing company behind this, we drove many different transformations over the last years. He's a special Cisco partner. The company is special for me because from this company, I bought my first car. My actual car is also from this company. It's a Ford Mustang. Now, please welcome with me on stage the Chief Information Security Officer of Ford, Patrick Milligan. That should feel a bit like home for you, because you've spent a lot of time on Cisco stages, on Cisco Advisory Boards in the past.
That's true. Yes. I've never gotten comfortable with this, but yeah, that's true.
Good. I think we don't need to explain to anybody who's Ford. I think that's an easy one. Maybe we start with you. Tell us a bit about your journey. How did you finally become a CISO?
Yeah. First, a little bit about the role of the CISO at Ford. It encapsulates all of what you would consider the normal parts of enterprise cybersecurity across identify, protect, detect, respond, recover. We also have our cyber defense capability and our security operations centers, which we run ourselves. Under our remit is vehicle and product cybersecurity. What goes into our vehicles, as well as all of our products and APIs, our manufacturing cybersecurity, our OT environments, our industrial controls, things we consider critical infrastructure, as well as third-party risk management and working with our supply base to ensure supply chain security, et cetera. I would like to say I know a lot of the Ford teams out there in a lot of those spaces, not only in cybersecurity, but across all of the services we deliver and work with.
I want to thank you guys for everything you do. The journey to the CISO role for me personally, it absolutely was not a destination. This was not a role I thought I was going to wind up in. I'm in my fourth decade with the company, which I still can't really wrap my head around. I started out as an engineer in one of our production facilities. I have been in the production environment, manufacturing, engineering, product development, and IT. I would say the last 25 years has been very focused in roles around cybersecurity, our vehicle software and electronics, working with connectivity both inside our vehicle, security inside the vehicle, connectivity outside of it. A lot of roles in infrastructure technology, dealing with engineering and operations of our network, compute, storage, facilities, security. Just a lot of opportunity and experience around technology and security.
That led to being asked to step back into the CISO role in 2020 for various reasons. It's my second go around doing it. So really been in the role six years in the present instance, but 10 years overall, great role, even greater company, just a privilege to work there.
The role of the CISO, there's a lot of discussions. Should this leader report rather to the CFO, or should he or she report to the CIO? What is your view and your take on this one?
There is no across industry, and we talk about this all the time, there is no silver bullet of where the CISO should report. If there was a single answer, we'd all be organized the same way. We have already figured this out. You can find CISOs reporting anywhere in the organization. Within Ford, it has always reported to the CIO or within the CIO's organization. At present, I report directly to our CIO, who is probably more knowledgeable and better at cybersecurity than I am, which makes for just a great working environment. It's very empowering. A lot of accountability comes with it. It really is phenomenal to have Mike Amend, our CIO, to have his influence with the rest of the Executive leadership team and help us to deliver cybersecurity to enable the business, not just for security itself.
I personally would find it hard to imagine what it would be like in this role of a CISO without having just been totally integrated with the entire IT organization because of what it takes to deliver these services and solutions.
I remember, I don't know if it's two years ago at Cisco Live, where you were kind of telling us, "Bring networking and security closer together.
Absolutely.
Create secure networking. I hope you can see we delivered on this one, Jeetu and the team.
For sure.
Tell us a bit more about your long-term partnership with Cisco and what makes it a special relation.
Yeah. Hopefully, I can do justice to this. The Ford-Cisco relationship started from day one. I do not know if a lot of people know that. We bought some of the very first routers from Cisco. Really, a 40-year relationship with the company. I started getting deeply involved with Cisco in 2009 when I moved into the infrastructure technology space. First thing that I found myself involved in was just an infrastructure refresh, not necessarily something big and sexy.
It was my first exposure to the innovation and assistance that Cisco can bring to companies such as Ford, not only on the technology side, but on the business side, just what we had to go through in order to prepare to refresh all these assets across our whole global footprint, whether that be the partnership we established with Cisco Capital and leasing these assets, the engineering expertise Cisco brought to the table to help us figure out how to go refresh our whole global footprint to refresh our infrastructure. That established a level of trust with me with Cisco and what you guys brought from a day-one and day-two perspective. Immediately behind that, we did a data center consolidation. We had six global data centers, consolidated them down to two in North America.
As part of that, we worked with Cisco on basically we run a private network inside Ford. We re-architected, re-engineered, redesigned our entire Wide Area N etwork, including our regional network hubs to establish connectivity back to these global data centers, introduced WAN acceleration at the time, and basically just dealt with transport optimization across the whole company. Our Webex journey with you guys, whether that be audio, video, or data conferencing, and then building out our next-generation workspaces with Cisco conferencing technology was behind that. We wound up building two enterprise-class data centers from the ground up. These are world-class facilities. In the consolidation, we worked with Cisco to build out our whole global network. Building these two new data centers allowed us to design and engineer a network within the data centers themselves to deliver a world-class compute environment inside of those working with you guys.
Expanding into the vehicle space, not only vehicle connectivity, but introducing Webex in our vehicles.
I saw this.
Yeah. Jim Farley, our CEO, was on stage with you guys here at Cisco Live a few years ago talking about that. I mean, who knows where this is going to go?
Before we jump to the AI one, you heard all the announcements and saw where we're moving. Just check, is this serving your needs? Is this meeting what you expected from us? Is it helping you to manage your complexity?
Yeah. Absolutely. Going back, you referenced the comment from a few years ago around networking and security. All of this, it comes together. Whatever version of AI you're dealing with, the scope of where this stuff interacts within our company across the whole network, tying network and security together is just fundamental to the success of us being able to not only enable the business with the capabilities they need, but from a cybersecurity perspective, enable these capabilities in a safe and secure way for the company. You guys are absolutely heading in the right direction.
Perfect. Great to hear. Now, talking about AI, AI for a company of this size, I assume it's a challenge and a huge opportunity. How are you, in general, approaching the usage of AI?
Yeah. A lot of things go into AI. As an automotive industry, obviously, Ford has identified our own high-value use cases that we're investing heavily in and going after those. Other OEMs are doing the same thing, I'm sure. Our technology partners, providers, suppliers are also going to be building out these capabilities, and Ford's going to want to take advantage of them. Other companies are going to want to take advantage of them. The opportunity or challenge within AI from a personal perspective, it's not just figuring out how to enable and secure the agents we're going to build ourselves, but also the ones that we're going to be leveraging from our industry partners, suppliers, and providers. Highly challenging environment when you deal with that.
I will say from a Ford Motor Company perspective, when we start dealing with those challenges, the principles and ethics associated with responsible use of AI are very important to us. Protecting our customers' data, ensuring we're engineering in privacy by design and effectively everything we do is very important to us. Establishing trust with our customer base and delivering mobile solutions and capabilities to them that enrich their lives is very important to us. Data security, data privacy, data protection, very, very top of mind, especially when you're dealing with LLMs and the immense data that's in those environments. Right behind that for me is robust identity and access management solutions. Not that we don't have the technology in place to help us do this today, but the scale we're going to have to deal with is just exploding.
I understand the viewpoints of human in the loop of all this stuff, but I'm telling you, the threat actors who are out there are not concerned with humans in the loop. They are giving these Agentic AIs a goal and turning them loose. This whole environment of machine IDs, we're already at a ratio of some people would tell you 100 to 1 for machine IDs to human IDs. These machine IDs are going to be different going forward. It's not a set, "Hey, I'm going to give you a specific task. Go do this over and over and over again." Very easy to control, very easy to secure. These are going to be machine IDs that have human behavior associated with them, whether it be an Agentic AI that has a machine ID that's going off and doing who knows what.
We have to have the capabilities in place. We have to get in front of this. Again, not necessarily a technology shortcoming, but we have to get in front of this to be able to manage these IDs, provide the appropriate authentication, authorization to do this in a safe and secure manner.
Getting in front of this, there was a couple of questions that came up in different sessions about how far do you control. I asked Vince this already. Most CIOs that I talk to say, "Oh, they don't know what is really happening right now. Who's using what type of agent?" Being the CISO, I think they will come to you at the end and say, "Hey, Patrick, why were we not able to stop this or make this not happening?" How you balance, again, out the challenge of allowing your teams across all of Ford to utilize and embrace the power, but on the other side, keep safety and security on the standards you defined?
Yeah. I like the way you worded that in terms of balancing it. This is the single biggest challenge, I think, almost all of us have in dealing with this. We do not want to stop the usage and the value-added proliferation of this capability anywhere. It is what is going to enable Ford Motor Company. It is what is going to enable your companies. It is just this is where this is going. It is here. It is how do we enable it? Because you are right. Everybody is going to come to us saying, "What are we doing to ensure we are doing this safely and securely?" We do not have the answer to this today. I have made a comment to people before that if somebody tells you they have this figured out, they are lying. We do not have the answer to this.
It is something we all have to work together to figure out how to do this in a manner that kind of like all your guys' announcements you're coming out with, going down the path of how you securely operate and enable AI and all the technologies that come behind it. It's just this is a cross-industry problem that we all have to work together to solve.
I can tell you, we have something cool coming up there that should help you on this one. Before we run completely over, just maybe as kind of my last question, share a bit your experience with Cisco and also maybe any recommendation and outlook into the future you want to give to all the amazing IT leaders in the room.
Yeah. In terms of the relationship with Cisco, you know I'm going to tell a story. I hope I don't embarrass anybody out in the audience. I mentioned we did this whole data center consolidation, just a major change to our global network. You can imagine a multi-year event, hundreds, if not thousands of deliverables specifically in the network space. Everything's important in a program and project like that, but there are those certain events that come up that are just defining moments for how your program or project goes. We had one significant, very significant weekend network change that had to happen that was going to affect our whole global network. This is one of those things that can go upside down very quickly and get you in a lot of trouble.
This was something that literally had to come off at this particular point in time. We had plan Bs and backup plans and all that kind of stuff, but that would have been a nightmare for the company in terms of what we had to do to deliver this network redesign. This is something that had to happen. The Ford and Cisco network team working on this together on the given weekend all squirreled away in one of our data centers. As luck would have it, there was just a major storm in the area. They all knew it was coming. In the midst of this event that just absolutely has to go off, the tornado sirens went off, tornadoes on the ground. I ask you, what would you do in that situation?
I'm not going to name the names because some of them might be out there. I'm not a proponent of what they did, but this team started pulling together just hundreds and hundreds of feet of power cabling, network cabling, picked up all their gear, got all the kit, and basically re-established their command center in the interior men's room of our data center in the midst of a tornado and delivered this network change. I don't forget stuff like that. When you ask yourself, who do you want working with you? Who do you trust? That's what excites me about this relationship and what we can do together. Who knows where things like AI are going to go, but what we do together is what gives us the opportunity to do great things. Advice to Cisco?
Man, keep doing the things you all do at a level that people can't just ignore you. Deliver network security capabilities that enable all our jobs that nobody else can do.
Perfect. There's nothing to add. Big applause. Patrick Milligan. Wonderful. Thanks a lot. Kerry, I think we need to check our story because, first of all, digital resilience includes tornadoes, and future-proofed workplaces need to include men's restrooms. Thanks again to our two great customers. I hope you felt well represented on stage seeing how we are focusing on you. Before I hand it over to Chuck to wrap it up, let me just be very clear. What is my mission as Cisco's Chief Sales Officer? My job is to ensure that you get the most out of our solutions and services. That's my number one priority. Together with the 90,000 people of Cisco and all our amazing partners out in the field, I'm committed to help you to ensure that it's easy to work with Cisco and that you get the value you expect from us.
This is for today and for tomorrow. It's a great honor to serve you as a customer and to help you to deliver on your mission. Big thanks for being here with us at Cisco Live. Big thanks for your trust, and big thanks for being a Cisco customer. Thanks a lot. With this, it's a great pleasure to welcome back on stage our Chair and Chief Executive Officer. Please welcome on stage Chuck Robbins.
Thank you, Robb. I want to thank Vince and Patrick. Thank you for that. That was really, really, really good. Give them a big thank you. Have you guys had a good first couple of days?
Yes.
It's good? No, we started off yesterday with just a plethora of innovation from the team. Continued with more innovation from the CX team today with Harrison Ford. Thank you for coming. We have to figure out how the anti-aging thing is working. Thank you for your partnership, and thanks for being here. The customer discussions were fantastic. I just have one question for you. Did we meet your expectations relative to the amount of innovation that we delivered this week? Yeah? I do not have a lot to say other than, obviously, we still have another day of the event. I primarily wanted to just say thank you and reiterate what Oliver said. I want to thank you for your partnership with us. Thank you for being a partner. Thank you for being a customer.
Thank you for your time because we all know that that is the most precious thing that we have. For you to choose to spend as much time with us as you have does not go unnoticed and certainly is not underappreciated. We greatly appreciate you being here with us. We look forward to actually helping all of you take advantage, hopefully, of a lot of the innovation that the teams have announced over the last couple of days. DJ and Anan, I think there was some slight interest in AI Canvas. I heard that a couple of times. We look forward to spending the next year helping you achieve what you are trying to achieve and do the things that you are trying to do as you continue down your AI journey. Thanks for being with us. Enjoy the next day. Enjoy the concert tonight.
We will see you soon. Thank you.
Welcome, welcome back, everyone. Oh my gosh. Do you feel like me and just felt all of that great content and all the innovation? I mean, just starting with Liz Centoni and all of the great things that we're doing for the Customer Experience, it's real. It's coming. It's not something that's in the future, way out there. We are leveraging AI for real-world impacts. I know how it feels being opening up a case and wondering, like, hey, is there something else I should know about? I know customers, how you feel when you have these devices in your network, but you don't know if there's something maybe you should be upgrading. Is there some issue going on? Can you be proactive in your defense? Yes, you can with Cisco. I love that. I was just like smiling the whole time. I also loved hearing from our customers.
There's nothing like a real story. What we're doing at the University of California, San Diego, with our solutions, what we're doing with Ford, powering the Automotive industry, one car at a time, really, I thought it was incredible. I especially loved how we were infusing security directly into the plant. That part, because my security background, I was just like, yes, yes, security. Like Jeetu said at a point, right, we know that security is accelerating what we do. It should not be a hindrance. All of this is just great news. Super super great news. Robb, I know you're out there too. What were your highlights or how are you feeling right now? Because I'm up here. Where are you at?
I've been out there, and I'm right behind you. We were on the keynote floor together because I'd kind of forgotten that until I'm like, wait a second. I feel like Lauren's presence is nearby. Sure enough, there you were. To me, the thing is, especially so you mentioned Liz Centoni, and I like where we're now making real from a CX experience, the integration of AI, not just in terms of natural language interaction and such like this, but really saying this notion of Agentic AI and how can we use this to take some of these, what's the right word, just long-term challenges we've always had where customers have so much distribution, so many assets, both hardware, software, and sometimes somewhere in between, some things from the cloud.
How do we get a complete view of not only where those are at any given moment in time, but what release are they at? What's the situation with patching? How fast can we do that? We're announcing abilities to kind of, we'll talk about this a little bit further, micro-segmentation on the fly to kind of buy yourself more time as we shrink that window of exposure from a security perspective. I like what we're doing in CX because I think we're simplifying, making it much, much easier for customers to keep on top of just what is a lot of very important, highly distributed gear. Yeah.
Yes, yes, yes. We have some really cool things coming up next too, right? Like I've already seen on the agenda, which you all are probably doing as well, some quantum innovation, how we're securing networks with that. I know that's going to be on my agenda to make sure I go look at. We got some things we're doing with our hybrid mesh firewalls, a lot of cool content. Stay with us. Don't go away. We will be here right after this.
Hey, Elizabeth. Still there? Okay. Change everything. No problem. Where are we going now? The perfect getaway for my sister and I. Skiing, wine tasting, spa. Sure. Sounds like a great time. I'll get right on that. I found a number of vacation packages available in your time frame. What day would you like to? Wait. I've got it. Miami. That's the destination. Of course. I found several flights that fit your preference. Window, eighth row, left aisle. You're amazing. Thank you. I try to be my. Actually, what about Greece?
Introducing Webex AI Agent. What's my account balance? Is this covered by my employee benefits? I'd like to exchange this for a different size, and color, and style. Where hold times end and self-service begins. Meet AI receptionist for Webex Calling. You're always on. Virtual receptionist that can answer questions, book appointments, and transfer calls so your business can boost customer satisfaction without adding staff.
Thanks for calling Florissa Dental. I'm your AI receptionist. How can I help you today?
Hi. Yes, I have a toothache, and I'd like to schedule an appointment. First, I want to make sure that you accept US insurance.
Yes, we do. U.S. insurance is covered 100%. Would you like to speak to an insurance specialist, or are you ready to schedule an appointment?
I'd like to go ahead and schedule an appointment. Thank you.
Great. I'll transfer you to a nurse who will give you an initial consultation and help to schedule your appointment.
Thank you.
Deliver outstanding customer service. AI receptionist, available later this year as an add-on to Webex Calling. Best Customer Experience is the one your customers never have to think about.
Hi, Sam. Sorry about your train cancellation. Let's get you booked on the next available train. Because you communicate proactively, anticipating their needs before they become needs. Because you deliver 24/7, always-on service with AI agents that offer personalized experiences. Because you can scale with Agentic AI that completes tasks and resolves customer issues in the languages and channels customers prefer. Because you empower customer-facing teams with AI tools so they can stay engaged while providing higher-touch support. Because you harness AI to turn insights into action and enhance every part of the customer journey. Experiences so effortless and engaging, your customers keep coming back. That is the power of Webex and AI-first Customer Experience.
Four years ago, we wanted to change two things. We wanted to change the way we worked, and we wanted to change the sport of golf forever. The solution of both was Webex.
My name is Cameron Pollard, and I play golf. When I first started competing, playing in front of people wasn't my thing. Being autistic, it was hard at the start. Then came the Webex Play Series.
The idea was one tournament, one course, one title, one prize purse, men, women, and all abilities together. Golf is a game that's hardly changed in hundreds of years, and it took some doing. Webex was showing things that we didn't know we could do in our sport, and it enabled us to connect our athletes all over the world to get our athletes ready for the global game. It is working for them. We've seen athletes such as Grace Kim start in the Webex Play Series to go on and win on the LPGA.
I do want to show them that you're actually good and you belong there. I'm not just a bloke with autism. It's just because we're good enough to be there. It's a pretty awesome feeling.
Since the inception of the Webex Play Series, female participation has grown 25%. Participation in general over the last year has grown from 3.4 million to 3.8 million golfers in Australia. That is phenomenal. As for the team, Webex changed the way we worked, and Webex's AI has been amazing. People focus is a great feature for us. It allows for everyone in the meeting to feel included and be seen and heard on the one screen. Real-time translation is very valuable for us at the PGA. We talk with global tours on a regular basis. Things used to take twice as long with our team out in the field. We began this journey thinking two things. We would change the way we work, and we would change the sport of golf. What we never imagined is that we'd actually be changing people's lives.
Cisco and SAP have agreed on ambitious milestones for the migration of our more than 250 offices globally. Right now, we are planning on having upwards of 350,000 devices on the network, including IoT devices. We have more than 100,000 employees. They need best-in-class IT infrastructure to work wherever, whenever, and however they choose with the devices they prefer. With Cisco, we've made our network simpler and more secure. Cisco SDA also enabled us to improve our zero-trust access policy, ensuring only trusted devices can access corporate resources. With Cisco Spaces, we could leverage this foundation and establish a global platform for smart offices. Our first experience confirms our hopes and expectations that we can manage everything in one place together with a professional partner like Cisco.
Tribeca has always been curious about technology. I feel that so much of Webex and Cisco specifically have done for us is help us lower the barrier of entry so that all artists can participate and be involved in the festival.
We're using Webex events in order to make the experience of De Niro Con much more interesting, immersive, and connected for our audience. A frictionless experience was key. We don't want our fans and customers worrying about how to find things or where to find things. All we want them to worry about is how do they have the best possible experience. We want to continue to tap into the broader suite of capabilities that Webex has, including a lot of the AI capabilities that they continue to build out.
It's important to work with companies that have shared values because it's a partnership. It helps us amplify our voice. To a certain extent, it's a shared mission of Cisco to be able to amplify diverse storytelling. You need to have global partners if you're a global festival. We are truly honored and proud to be partnered with Cisco.
Webex Calling Customer Assist enables your local and regional employees to deliver seamless customer assistance with AI-powered features in the collaboration app they already use. Click to Call allows your customers to get in touch with one click. No login required. Once they've reached out, any team member can answer and help. With AI Assistant, they'll have efficient tools at their fingertips, like real-time call summaries that can be shared when transferring a call. With the Call Sentiment for Supervisors feature, supervisors can assist team members whenever they need support. Enable your local and regional employees to deliver seamless customer assistance empowered with AI in the collaboration app they already use. Webex Calling Customer Assist.
Thank you for calling. You're caller number 25. Your estimated wait time is 37 minutes. Hi. I'm Blossom, an AI agent. Hey, how are you? For your rush delivery, I can help you find the flower you were looking for. We have. Yeah. It's my grandma's birthday, and I need her favorite flower, but I have no idea what it's called. It's bright yellow with, like, bell-shaped center, very thin stem. Sounds like a special occasion. Okay. Based on your description and our stock, I would recommend daffodils. Oh, I want a dozen, please. Understood. A dozen. Coming right up. Okay. I think we have what you're asking for. I've sent it directly to your phone. Yes. The daffodils are perfect. Can I order a bottle of champagne as well? That sounds delightful. Unfortunately, we do not offer alcohol purchases online, only in store. However, we do have artisan vases. Would you like to add one to your order? That sounds good. You got it. If you're ready to complete your purchase, I can text you the payment link. Yes, please. Let's get that payment arranged. I've sent a text message with the confirmation. Have a great day.
We are back. We are back. We are back, everyone, with two exciting guests with none other than my lovely co-host, Robb. I am going to actually let you go see this excitement that we have in store for you. Passing over to you, Robb.
Absolutely. Thank you so much, Lauren. Yes, I'm excited. I've got someone who's actually become a friend, Giles Adams, CEO of VQ Communications.
Yeah. Good to see you.
Helen Briggs with a collaboration for EMEA, see how the brain just stops down when it comes to targets?
Europe, Middle East, and Africa.
There you go. She knows where she's from and what she represents. That's perfect. Just before we get into the questions, I wonder, first, could you explain what you're responsible for, Helen? Then we'll come to you, and just kind of give us the quick on who VQ is and how long you've been part of Cisco because you guys are a very close part. Let's see what you have to say about that, Helen. Tell us what you do.
Yeah. Thank you. I actually run the EMEA business for our collaboration architecture. Collaboration, as you've heard by the keynotes, is really key across both Customer Experience and employee experience. We're responsible for delivering that into all of our customers and working really closely with our partners as well.
Absolutely. Very good. Giles, I'm curious. VQ is a video conferencing provider.
Yes. Not quite. We've been developing software to enable people to build services for coming up 20 years. I am in a very fortunate position, currently being the CEO. We've got a great talent pool of engineers. They're all based in the U.K. My job is to provide some guidance and get out of the way and let them do their clever things.
Specifically, though, you guys are building your solutions. You're using Cisco gear 100% across the board, but you're doing it in such a way that is not something Cisco does as much of historically. You have become a very crucial part of the relationship for many customers who have mission-critical networks. Can you explain, if I'm setting you up correctly, can you explain what I mean by that?
The marketplace that we address has been always around people delivering mission-critical services. That typically requires really stringent, high-quality software for people to build a conferencing service, which may be in a government agency where people are looking to secure their nation's secrets. Increasingly, we're seeing that across enterprise customers as well. In the relationship with Cisco, which has been direct now for four or five years, we've done 500 or so projects together during that time. We jointly interact with key Cisco infrastructure and add something to that environment.
Okay. This would be everything. This is key communications infrastructure specifically.
Right.
I'm curious from the market perspective, Helen, what you're seeing, especially in EMEAR or anywhere you're comfortable speaking to. Obviously, Cisco has been moving to cloud. Everything goes to cloud. For good reason, not just because some things become easier, but it becomes a better way to do scale as we get better and better at this. Obviously, we do not see that stopping, but we do have many customers that are concerned. They've always been concerned about, "Where is my data located?" That continues to bubble to the surface. I think it's given new energy to doing more stuff on-premises. We're not always engineered for that. I'm curious, where do you see the market going on cloud versus prem versus hybrid situations?
Yeah.
What are your thoughts?
Cisco is cloud-first. It's not cloud-only. That is to really cater to all of our customers globally. Whether you're in a certain segmentation that's very highly fundamentally flawed by regulations or mandates, or whether you are able to actually have some applications that can actually be offshored, so outsourced and cloud, or whether you actually have them inside, we actually cater for all. Just to actually be really clear, an on-prem scenario is really about where that infrastructure lives. It's where it lives and also where it's maintained. There are some real positives with that. It sits in a customer's data center. The customer themselves, they maintain, they secure, they manage, and they actually deliver software updates, which means that they're in control. It also means it's quite heavy on resource and IT.
Now, you flip that over, and from a cloud perspective, then Cisco takes all of that complexity away via our Webex offering with video conferencing. With that, it basically means that we can actually offer at scale agile innovations such as AI, such as collaboration tools. As Giles did mention, we have to also interact with the customer scenario and the customer's environment. They may have certain back-end applications or tools that we have to interact with from a cloud offering. That is where VQ comes in, of really understanding that customer's infrastructure and being able to be flexible. Some applications will be cloud, some applications might be on-prem. It makes it a very complex nature for an in-house IT department. That is why Cisco works with VQ.
Yeah.
Yeah. VQ certainly works with cloud just fine. You have developed tools, continue to develop, because one of the challenges when you start trying to remain distinct or have the ability to draw the line on where the cloud begins and ends in your operation, wherever the customer wants it, you also potentially maybe lose some of the good things about cloud and some of the services it delivers. You guys have consistently rolled out technologies, as I understand it, to bridge that gap so that customers can have the best of both worlds. Is that true?
Yeah, absolutely. I think trying to narrow the gap, because as Helen rightly points out, there is massively more compute available in a cloud environment than there is in an on-premise environment, which may be air-gapped and totally locked down. What we've always been trying to do is to make sure that there isn't a massive difference in the services that people can experience. What we don't want is somebody perhaps who works in a government agency going into a secure environment, having a conferencing experience, and then perhaps going home and then using another experience and going, "Well, they are completely different." You don't have two different types of phone experience. You have a phone experience. Being able to add back functionality that people just expect to use the term table stakes is what we've been trying to do and continue to do.
Of course, one of those things is, in a very simplistic view, adding back things like some of the AI capabilities, like meeting transcription, translation, all those types of things. That is something that we're working on quite heavily at the moment.
Which is funny because we think, "Oh, well, that makes a lot of sense." For you, in the actual implementation of that, your engineers are figuring out, "Well, that means localized models.
Absolutely.
There is a shift and an expectation and probably a lot of pre-work that goes into that. Because I think many customers may make the mistake of thinking, "If I want to take more of this under my control, why can't I just do it myself?" You're like, "Well, you're welcome to try. Here's my number.
Yeah. Yeah.
When you're ready to get serious.
It's actually quite complicated delivering services in an air-gapped environment. All of those reliances that people are familiar with in a typical IT deployment where you access, for example, licensing permissions through a cloud-based service just don't exist in an air-gapped environment. There is quite a lot of complexity. That is why it's quite a niche environment in terms of the special engineers that work in this space. That is why we've got this great relationship with Cisco because we've been partnered together for a long time. Some of the customers, which obviously neither of us can share in an open environment, are obviously fairly paranoid about our services.
I want to ask paranoid questions, but I get it.
It is really important. We leverage the best infrastructure. Cisco and VQ dominate this space. There is nobody else in this marketplace. There are other vendors that will talk the talk. When it comes to getting into this type of environment, delivering high-volume services for customers right the way up to, if you could call it the sort of pyramid of paranoia, if you go further and further up a particular environment, it has to be resilient. It has to just work all the time and every time.
I know you guys have taught us a lot about how this can work over the years because this is not a new relationship. I wonder, I'm going to start with you, but I want you to piggyback off this from your perspective. When did the Cisco relationship start? Are there some key moments in terms of how, because I feel like a lot of the Cisco engineers, from my experience, look at VQ engineers as just another department almost because you're working towards the same goals, definitely separate companies. We've been working tightly together for a while.
Yeah. So we've had a direct commercial relationship with Cisco for four and a half years. There's a program called Solutions Plus. This is where Cisco's recognized, and it's not just VQ, that there are gaps in the Cisco portfolio, which is not relevant for them to invest in because either it's too small or it's too niche in terms of the special skills that are required. Cisco has this program called Solutions Plus. As I say, we've done some 500+ projects together. The pipeline is growing, not just in America. It's a real global program. It works really, really well. It's really where, over time, the recognized skill set that we have and the software that we have is solving real-world customer problems. That's the key thing, making sure that we continue to deliver to that.
Yeah. Helen, I think a lot of times these guys are very well engaged with our defense customers. Anybody that's dealing with sensitive data on a regular basis, and they probably know each other on a first-name basis, what is your impression of those relationships and how important it is to Cisco?
Yeah. Dynamic changes are always happening in the market. Where we are today, we're seeing that. Actually, with the on-premise that we've been speaking about, that's actually growing in certain markets and certain verticals. Giles and I were talking last week because I went to see a customer in Northern Europe. It was around the county courts. Some of the stats that they're actually trying to strive through at the moment is that one in every three court appearances now is on video conferencing. Before the pandemic, it was quite low. The pandemic brought more so. People now see it way more flexible as part of their daytime. With that, it becomes one in every five sessions need an interpreter as well. Due to the sensitivity of that call, 155,000 events are actually recorded and stored on an annual basis.
It's huge in type to that dynamic. With them, they have to keep that very, very safe. They have to keep that away from even some of their geo partners in country, whether that's the police forces or other areas that they actually have to contact. They will have certain parts of those particular interactions that can be shared. That's when VQ can come in and really start to prioritize what the data sensitivity is and the priorities are and how we can actually start to move that around to make that ecosystem work. We were just talking about that last week. It's integral that this partnership for our customers just keeps growing and growing.
Yeah. What I'm hearing is, and we're going to wrap things up, what I'm hearing is it's really Cisco, obviously, this week, as we always do, we're talking about data being so important and also differentiating for all of our customers. They need control of their data. Sometimes it's us giving them the tools to see where their data is and how to interact with it, certainly to be able to move it around at the speed and with the latency that they need and to communicate with it, but always knowing at any given moment that they're in control of it. It feels like you guys are working tightly together to be able to do that. Thank you for taking the time. I hope the conference remains good for you.
Thank you.
Your booth has been very busy downstairs. It's nice to meet you as well, Helen. With that, I'm going to go back to Lauren.
Hello. That was really insightful. My favorite part was really it doesn't matter if you're self-hosted in the cloud. You have a video experience that's the same. VQ and Cisco partnering together to give us that. Now, I'm seeing in my peripheral, like Steve was saying earlier, I see my lovely friend, Michelle, over there with a special guest. Michelle, can you go ahead and introduce us to your guest there?
Lauren and everybody watching, welcome back to the show floor. I have a new friend, Arpitha Shetty, who is a Product Lead of the Secure Unified Branch, which is visually right behind us. I want to ask her a couple of questions, starting out with, give us your elevator pitch.
First, I want to say thank you, Emma, for talking to me today. I'm really excited to be talking about Unified Branch. My 60-second, 30-second elevator pitch would be Unified Branch is a platform approach to managing our three key networking components, which is routing, switching, and wireless, all under a single management platform, which is the Meraki dashboard. You know the cloud dashboard to be intuitive. What we want to give to customers is streamlined operations. We want them to be focused on what value they are generating for their organizations instead of worrying about disparate tools and technologies. That is why we are excited to be announcing Unified Branch.
I love it. I love it. That resonates. That is a strong message. What sets this apart from things that we have done in the past?
I'll tell you what sets it apart. Before that, I may have to tell you a little more about Unified Branch.
Do tell.
Yeah. Unified Branch has a few core critical components, which is first, the platform that we talked about, cloud-delivered, intuitive, single pane of glass. That connects to our core technology, which is routing, switching, and wireless. What we have added to Unified Branch is a concept of Cisco-validated design, also known as CVD. CVDs are not new. They've been around in Cisco for a while. What they are are Cisco's best practices. Imagine 40 years of Cisco's expertise all documented and provided to customers to tell them what the best practices are to deploy a service. What we are doing with CVDs is now we are delivering a CVD for the entire Unified Branch versus just for a solution. We are not just giving you a document. Customers don't just want a solution. They don't just want features. They want validation.
They want confidence. With our CVDs, because they are thoroughly tested, we are giving customers the peace of mind to deploy our solution. That is something that's brand new. Assurance is important. Insights are important because customers are moving more to cloud-delivered applications. We are leveraging ThousandEyes, which is a really cool technology from Cisco. We are using ThousandEyes forecast to give customers visibility into their networks. That is a part of Unified Branch. The main thing is branch's code. We have given customers a simple, intuitive dashboard to manage their branches. What if they want more automation? What if they want to leverage DevOps principles like version control, automation, testing? For those customers, we have introduced branch's code.
We're just trying to make things simpler for customers to adopt all our technologies, to adopt all the cool technologies that Cisco develops. That is Unified Branch.
That was incredible. I just learned a lot from her. Okay. One more question. I'm going to wrap it up. How are customers changing the way they think about branch management?
Like I was saying, customers want validated designs. Customers want peace of mind. Today, customers are facing problems with IT shortages. There is a talent shortage. There are disparate tools that they have to manage. They want something that is simple and intuitive. That is what we think is changing with customers today. They want more automation. You heard the keynote yesterday. It is all about AI. I think that is what is changing with customers today. We are creating technology and products that help customers move to the next digital transformation.
That was incredible. I can't stop saying I learned a lot from her. Thank you so much. I'm sending it back to you, Lauren, in the studio.
Thank you. Thank you. I learned a lot as well. We have them coming. We have another interview. I'm passing it back over to Robb to hear more.
You got it. Thank you so much, Lauren. This is fun. Craig, I met you when you were out hyping Hypershield. Hyping Hypershield. I did not think about that. Yeah. Still worth hyping, if you will. Secure Mesh Firewall is what I wanted to talk to you about and understand, kind of, am I saying that right? Hybrid Mesh Firewall. It should be Secure. The Secure is assumed when we say firewall. Can you tell me what that means and how unique is that naming in the market? Because it was a little bit more known than I had even realized.
Yeah. Hybrid Mesh Firewall is a term that Gartner coined to talk about the convergence of management of different types of firewalls. We have three types of firewalls traditionally that exist in an enterprise. We have traditional box-based firewalls that run on-prem. We have cloud-native firewalls that run inside of hyperscalers. We have firewalls as service that runs as part of our SSE or SASE solution. Gartner wanted to push the industry to converge on a single enforcement stack, a single management stack for those three firewalls. Now, we've taken it and expanded that definition.
When you and I were talking a little bit earlier, and this is something I wish was sometimes made more clear to everyone because it's what I struggle with, is when we change the name of things. There's been some significant name changes in security, but there's a good reason for it. I wanted to can you kind of tell us what the translation is from what we used to call it and what we call it now and why?
Yeah. So what we did is we tried to rename products to make them easy to understand. So we used to have.
A functional name.
A lot of, yeah, a lot of bespoke names, a lot of acronyms. It was like a secret jargon to understand what all the Cisco products did.
A secret to my power.
If you look at the elements of our Hybrid Mesh Firewall, two of the key elements are what we call today Secure Firewall. We used to call FTD or Firepower.
Firepower threat defense.
Yeah. You might not know what Firepower means.
No.
You might guess that it means firewall because it has fire in it. Secure Firewall, it's pretty obvious what Secure Firewall is. Similarly, we had a product called Tetration. You wouldn't know what that does. Now it's called Secure Workload. It should be very clear that at least its job is securing your workloads. It's trying to give functional names that make it easier for folks to understand what we do.
As we're adding new things, it becomes a little bit challenging at first, I feel like, to differentiate between what's new and what's just a name change. But there's new elements to all of these things. In fact, can you talk a little bit about because we've got increased enforcement points as part of the umbrella technology with this mesh.
Yes.
What are the elements that are delivering that for us?
Hybrid Mesh Firewall is a suite of enforcement points. The idea is a single policy that can be pushed down to a distributed set of enforcement points. Those enforcement points can be the Secure Firewall that I just mentioned, Secure Workload that I just mentioned, Hypershield, which runs in two form factors, either on the workload with eBPF, where it's powered by our ISO Valent Runtime Security technology.
ISO Valence, where we got the eBPF.
Correct o r inside our smart switch, where it runs as a network enforcer. We've added two new enforcement points at this show. One is Cisco ACI, which folks know, the data center fabric that they know and love. The other is our new Secure Routers actually have the ability to take security policies from Security Cloud Control using the same security stack and do enforcement right on the Cisco Secure Router.
When I think about this distributed nature that we're all kind of struggling with, the speed, of course, the distribution, and AI specifically, because we have AI that we're using to get more things done. You, as a security person, are constantly thinking about, well, wait, who's checked the AI? You guys are constantly thinking about both sides of that in all of this implementation, correct?
Yeah. I mean, AI brings a couple of interesting challenges. I always say Chuck talked about fusing security in the fabric of the network. I like to say we need to change security from being a place in the network we send traffic to being part of the network itself. That's really important when you think about AI because the speed that AI is operating at, you can't have a firewall box and all my AI traffic is going to go through that firewall box. You've seen those stats we have where a single AI rack has more traffic than the entire LTE network in the United States of America. I'm not going to put a firewall box in there. I need to distribute security everywhere.
By having the same security distributed in all these enforcement points, not only does it allow consistency, but it allows this horizontal scale of those security services so that as I'm rolling out these new inferencing capabilities, learning capabilities, other things with AI, and now agent-to-agent communication, I'm actually embedding the security close to where those things are happening, which allows me to not only have consistency but also scale.
I think it was in the G2 keynote he talked about kind of buying us time between, say, when a vulnerability is announced and the time in which a patch is released. Because we see threats come out much faster, of course, than the patch. You guys are the infrastructure, when deployed in this manner, has been given the ability to microsegment on the fly? Is that how it's working?
Yeah. Two things. One, I mean, for vulnerabilities and customer workloads, Hypershield is there to do autonomous segmentation on the fly and also what we call distributed exploit protection, where we can actually use eBPF to prevent a vulnerability from being exploited. This week at the show, we announced Cisco Live Protect, where we're taking the technology that does distributed exploit protection and bringing it to Cisco devices, starting with our Nexus switches, but it's going to extend to all Cisco devices. Not only a vulnerability in the customer workload, but what if Cisco ourselves has a vulnerability? We're going to allow you to, at runtime with no disruption, prevent that vulnerability from being exploited, which means you still have to patch the device, but you can do it at your next scheduled maintenance window. It's not a hair-on-fire emergency the way it is now.
It's not. It's just life in general. It feels like. I like that. You're buying some time. The first thing I thought of too sometimes is if you're locking that down, obviously, we have to be very intelligent how we lock that down because we're talking about production resources that are still in play. It's using them securely, but in some kind of a fashion that is still safe. I guess it's very exploit relevant.
It is. It's very targeted. What eBPF allows us to do is safely extend the functionality of the kernel at runtime and do it by extending an individual what we call hook point. That could be a single function call, a single system call, a single K probe in the kernel, which is the name for a hook point in the kernel. We are able to write very targeted rules that are actually executing as part of the kernel. Very fast, very safe. Imagine I'm not saying block this big thing. I'm saying this one function call cannot be called by this one process if it's running in this one network namespace. I'm being very, very fine-grained.
Something really a machine can only write that kind of rule on the fly, I would imagine, at that speed. I love that because that's exactly what we need to deal with today's realities. Craig Connors, thank you so much.
Thank you.
Appreciate you taking the time. I know you're busy running around this week explaining these concepts because security is a huge focus with everything that we're dealing with. With that, we'll go back to Lauren.
Hello, everyone. And all I heard was simplicity. I heard single pane of enforcement, single enforcement stack, actually. And it's music to my ears because if you've been in security long enough, you know security used to be an afterthought. But at Cisco, we are bringing it front and center. We're infusing it into everything we do in the policy engine everywhere. I'm super excited. Oh my gosh. Thank you. That just really got me really, really pumped up. Now we got a lot of good content coming over to you guys. Interview, interview, interview. Guess what's next? Another interview with Michelle. Michelle, are you out there?
Lauren, you know I'm here, ready with another interview. Thanks for bringing it back here. We're on the show floor. I'm here with Grant Shirk, who is in Product Marketing. Now, why I'm excited to talk to Grant is that I'm also in Product Marketing. We're about to geek out on the coolest job. What I want to talk to Grant about is security architecture. G2 brought this up earlier. What are the three pillars you want to highlight for people to see?
Yeah. I mean, if you go back to the conversation we were having yesterday, there's really three primary challenges that our customers are facing that they need solutions for at the end of the day. A lot of it is brought on by the advent of AI and Agentic AI. There are a ton of today challenges that are already impacting networks. Our customers really need three things. That's what we're here to talk about. One is delivering that operational simplicity across the network that is powered by AI as we're dealing with skills challenges and staff shortages and really just the complexity of the network overall. It is critical that we give our customers more and better tools like AI Canvas in order to collaborate, diagnose, and understand what's actually happening on their networks.
The second piece, though, is we talked about the need for more powerful, more intelligent infrastructure and hardware that's built for the pending demands that AI is going to bring. I still have that chart that we showed in the keynote yesterday of how traffic patterns are changing. We've gone from AI driving this kind of spiky intermittent traffic to now constant inferencing both at the data center and the edge. That is shifting the traffic that's flowing across the network, which is the second reason why we really need that different architecture to solve for these problems. The last one, which is really both a today and tomorrow problem, is that need for deeper, better, and more effective security both for the challenges of today. It is those three things. Operational security. It is hardware and infrastructure built for the AI era so we can be ready.
It is our ability to protect both the devices and the data that's flowing across that network.
Thank you. That was very powerful. I want to move it over to hardware for AI. Can we talk about the smart switches that we're introducing? Are they part of that?
We can absolutely talk about the smart switches. In many cases, they are the anchor in the infrastructure that's going to drive all of this. When you think about where all of that bandwidth, that connectivity, the need to scale out these tens of thousands of devices and the hundreds of thousands, if not millions of agents, you need a different kind of infrastructure in your switching layer. We are standing right now in the middle of the Future-Proof Workplaces showcase. We call this mission control kind of informally because this is how we're showing that unified platform coming together. Just over here, we actually have the new Cisco C9610 series smart switches. These are powered by a new class of Silicon One chips. They're actually pairing together both that Silicon One CPO and a second co-processor.
Their superpower is their ability to run both the networking flows and the security in parallel without impacting each other. For any of you out there who are still running, you might have a 6800 that's buried in a closet somewhere. I heard somebody talking about a Cisco 6000 chassis that had been drywalled in, and there was just one Ethernet cable coming out. Still running. Now is the time. Now is the time to think about.
This is your call to action.
Future-proofing that network. You're getting 10 slots. You can run two supervisors in parallel. There is some really powerful technology on this that's going to drive that post-quantum cryptography, protecting the data. Honestly, they're just really cool.
They are.
They're fully redundant with swappable fans on the backplane. These things are cool. I will tell you, while we're kind of off camera, I had to help carry one of these up a flight of stairs. It weighs 600 lbs fully loaded. It took five of us to get it up. I'm just really glad we didn't drop it.
How many flights of stairs?
Only one, but that was enough.
That's one too many. Last question. What else are you getting attendees asking about here at Mission Control?
I think there are probably three main questions. Certainly, one of the most fun things they're doing is enjoying the variety in the fixed access line of these smart switches, which is the C9350, which is just behind you. These are also Silicon One powered. They're available in 24-port, 48-port varieties. I think there's some capabilities, though, that we gloss over if we don't talk about if we're only talking about the security and other things. Each one of these, particularly the 48 ports, they're 90 W of PoE in every single port. You've got a full power budget. The power supplies, actually, if any of you have the chance, come over and play with the power supplies. They're titanium rated. They're incredibly sustainable and power efficient. You can do some amazing things with energy management.
As a matter of fact, right before we started this interview, one of the team over here in the center reminded me every single device in this section, from the Meraki MV cameras to the Thinlabs displays and the lighting, is PoE powered. It is all driven by the column in the center. This idea that energy networking is actually a real technology, proof of that is right here in the showcase.
That is such a great thing to end on. Thank you so much, Grant, for your time.
Absolutely.
Hopefully, you're not carrying anything else up the stairs today.
I will see what happens when it's time to take the place down.
That's true. No promises. Lauren, I'm sending it back to you in the studio. Thank you.
Thank you. Thank you. That was great. Now we are preparing for our next session. I know you all are just soaking in all this information, right? I hope you are. If you're not, you better be because we've got more stuff coming. The next thing we have is all about AI-powered assurance from the edge all the way to the cloud. If we think about it more and more, we're seeing devices that are managed, not managed, owned, not owned, all in our environments. It becomes even more difficult to really be able to pinpoint where the issue is that's occurring. Being able to find out the root cause, being able to understand what device, what team, if you will, what team is responsible for this issue is just getting harder.
As Cisco's blunt, we are building solutions to really help you navigate that, really be able to give you the end-to-end visibility that you need and that assurance. What's really cool is that we're bringing it as a single platform using things that you probably already heard of on the Cisco side, using ThousandEyes, some of those solutions. I'm a little biased, but Splunk, to help with the IT operations. I'm super excited to hear Sanjay and Diego really bring to life what we mean when we say AI-powered end-to-end. Come check out this session now. Splun
To explore how Cisco is reimagining network experiences with AI-powered management and assurance from the edge to the cloud and everywhere in between. You'll learn about innovations in unified management and ThousandEyes, the Cisco AI Assistant, and our integration with Splunk Plus. You'll hear a real-world Customer Experience. To get us started for the day, let's invite onto the stage Sanjay Mehta.
Thank you. Hi, everyone. Welcome to day three of Cisco Live. For some of us, day four. Hope it's all going really well for you. I'm Sanjay Mehta. I'm the Vice President of Product Marketing for our Networking Platform and Assurance. I've been in the operations space for about two decades: IT Operations, Network Operations, Security Operations. What's really, really key in this world is the complexity that's increasing. We've heard it all this week around AI and all these different factors. I've met many of you this week, a number of you at our customer Advisory Boards. You agree. You've said that complexity is increasing. It's real. It's growing. It's not getting any easier.
To help me later on in my talk, I'm going to have a fantastic customer, Itaú Unibanco, who's going to talk about how they're improving operations and using Cisco technology to do that. Again, welcome. Let's get into it. There's a number of factors that we deal with in operations and infrastructure all the time. In just the last two decades alone, we've seen the internet has really got us to rethink architectures and operations when we think about networks moving beyond the four walls of the enterprise. Then mobility meant that our devices are anywhere, not just on premises. With cloud, applications and workloads move off premises. We have to handle that. Each time, there's a point where we have to think about our architectures, our infrastructure, and crucially, our operations. We're in the midst of the new one, AI.
Many of us use AI every day, whether it's through search, Copilot, through an AI assistant. It introduces new demands on the network. With AI agents acting like humans talking to other AI agents at machine speed, we have a real demand on latency. If there's some issue with latency anywhere in the network, anywhere in the communications, it can impact critical workflows and break them. You've also got East, West traffic increases. You've got new sort of chains of experience that now exist at a high rate. All of these things put a demand on our infrastructure, get us to really think through and make sure that we have resilient operations. All of you, you are a rare group now. There's skill shortages out there. In our last customer Advisory Board, it was incredible. I would meet large global companies.
I could count the number of network engineers in any one of these companies on two hands. They were running global networks. It's actually incredible. At the same time, we know there's pressures on budgets, et cetera. There's that pressure to do more with less. What that means when it comes to traditional operations is it's challenging. At worst, they break. At best, it accelerates the timeline of usefulness in terms of your traditional tooling and the operations that you have. At Cisco, you've heard this week a lot about agentic operations. How we think about it in terms of operations is really operational simplicity powered by AI. What I want to go through are some ways to really help you drive operational simplicity with some ideas, with some technologies, all of which are available and we've announced this week.
We really want to get into it. It starts with our platforms. At Cisco, as you've heard this week, we're really bringing together our platforms. We're bringing together our Catalyst and Meraki hardware. It's now Cisco, Cisco smart switches, Cisco secure routers, Cisco wireless access points, one license across all of these, whether you're on-prem, in the cloud, hybrid, and unified management. When we click into this, there's really four areas. There's unified management across your Cisco and Meraki devices. There's assurance, this promise of delivery of experience from any point of origination to any point of consumption. There's APIs and integrations that bring all of this to life to give you extensibility on this platform. Then there's intelligence that's infused across all of it. Let's start with intelligence. Again, this week, we've heard a lot about agentic ops.
It's really the evolution of IT operations with an agent-first mindset. There are three core principles when we look at agentic operations. There's precision, that it's purpose-built. Our LLMs are purpose-built, for example, and it gives you that precision. It's based on data over and above a generalized LLM. That's a minimum bar. Secondly, that there are no gaps that you see across all domains, internal networks, your external networks, your IT security, et cetera. There are no gaps in any of that. Third, that it provides autonomy with oversight. There were some incredible talks this morning from customers in the keynote, if you were there, and also yesterday's keynote. It was really coming through that companies, organizations are figuring out what's the level of autonomy they're comfortable with. The point is we need to be able to set human boundaries.
We need to make sure that we provide humans in the loop. Because at the end of the day, it's all of you who are accountable. The agent isn't accountable. It's the humans that are accountable. We understand all of that as we're building out our infrastructure. This week, you've heard and really excited to announce the most advanced networking LLM in the industry and delivering really on those three core principles. Firstly, it's purpose-built for networking. It pulls decades of field expertise, troubleshooting paths all around networking-type scenarios. Secondly, it's using unique sort of trusted data over 40 years+ of Cisco with our Cisco U. courseware, with all of our incidents and all of that data, which means that, for example, when we've done our tests, it performs about 20% better than general LLMs for CCIE-type questions. That's really exciting.
Then it continuously learns with live telemetry, real-world Cisco tech and CX insights. It is always, always getting better over time. How does this all come to life? It comes to life with AI Canvas and AI Assistant. You saw the AI Canvas yesterday with DJ presenting that. It was fantastic. It is a generative UI interface that you can share with your colleagues and your teams, with other operations teams, IT ops, security ops, network ops. You can work from one place seamlessly. Also, with executives, you can ask questions. It will auto-generate visualizations, whether it is topology, whether it is charts and graphs. You can persist those over time for sharing, for future troubleshooting, for insights. It is truly spectacular. That is very exciting for us to see and bring to market. You will have that available in your hands very soon.
Our Cisco AI Assistant, this was in private. We've announced it before. It's now public beta. It provides you with that natural language interface to ask questions to help automate workflows. We'll get into both of these in a few slides, all powered by the deep network model. One of the core areas to underpin operations is management. Let's get into that. With our management, we're really excited to be bringing together the power of Catalyst with the simplicity of Meraki in our new Unified Management Dashboard based on Meraki dashboard. That gives you the ability to manage all of your Meraki devices, your Catalyst devices, as well as the new Cisco smart switches, secure routers, and wireless access points. It provides also the ability to manage your large campus environments. It has AI capabilities built in.
One of the other things we're announcing with our Cisco AI Assistant now being in public beta is new networking skills. For example, it starts with the ability to automate workflows for configuration changes, for switch migrations. I was talking to a customer yesterday who's talking about the human effort it takes to execute on repetitive tasks like switch migrations, for example, and was talking about how can we speed all of this up and remove error in the process, as well as new integrated assurance capabilities. For example, here, you can ask the question, "I want to replace my closet floor three with a new switch." It will go ahead and look at the from configuration to configuration, figure out the workflow. You can then test check it and say, "Does this look good?" Give the affirmative.
It will go ahead and execute on that. It does it fast, not hours, literally minutes, and obviously a lot less issues around errors because it is all well-defined, in sequence, done the right way. Really excited about this to be able to cut down on downtime and minimize errors. As I said before, our unified management works now across all Catalyst devices, wireless, switching, routing, as well as these large campus-type use cases for micro and macro segmentation, for standing up a new switch, for example. Also includes the cloud CLI because I know some of you still like interacting with your networks using the CLI. That is now available inside of the Meraki dashboard as well. Really excited about that. Now let's move into assurance.
Assurance is a critical part of operations too because it's how you really can deliver on the trust of an experience. At the end of the day, all of our infrastructure is there to deliver experiences, digital experiences. We know from earlier on that our networks now span both owned and unowned environments, as well as a ton of interdependencies there, CDN providers and DDoS mitigation providers, et cetera, et cetera. This is the new reality. This is the new world that we live in. At Cisco, we're delivering multi-layered assurance, giving you the ability to truly see end-to-end, borderless, from any point of origination to any point of consumption. It's not just about breadth. It's also about depth. You need the granularity. You need the telemetry.
You need the techniques to be able to see and understand, not just the breadth of that experience, but an experience-first view all the way down to the device. The three areas we've been focused on for this set of announcements this week are measuring everywhere that it matters, seeing every part of the path as if it was your own network, and then contextualizing the source. Because when there's a problem, it may be with a device, but it's really important to understand what services is this impacting? What areas of my enterprise is this impacting? What areas of the country is this impacting? What areas within my customer base is this impacting? We deliver that with ThousandEyes. ThousandEyes is how Cisco powers assurance inside of Cisco and for all of our customers. It's integrated everywhere.
Here's an example of assurance as it shows up through the Meraki dashboard. With ThousandEyes, we provide true end-to-end visibility from public clouds across the internet to endpoint devices anywhere, and that depth from a service, from an experience-first view across every layer of the network, from application service delivery all the way down to a specific device. It is integrated throughout Cisco. It is pre-integrated into all of the new Cisco devices that were announced this week. It is ready to go to give you that visibility that you need. With ThousandEyes, we also have AI Assistant as well. With that, you can ask questions like pick your favorite SaaS provider and ask the question, "What's wrong with my fill in the blank?" Whatever your SaaS provider is. It will go and look at pars.
It will pull data from your owned and unowned environments and give you links into topology views, et cetera. You can very quickly understand. A lot of this is done preemptively. It is done before users can impact issues. You can see problems before they impact your customers and move into a more proactive stance. Because it is generative as well, and natural language, it means that you can democratize access to this and make sure that more people can access this as needed through the enterprise. There are a few things we have been announcing this week that I want to get into. They really hit those three areas I mentioned earlier on. One is around deep visibility and measuring where it matters, the first of which is enhancing our traffic insights to give you the ability to actually do better capacity planning.
We talked about AI introducing new workloads onto your infrastructure. What if you could actually understand that and make sure that you could plan accordingly? We also are extending assurance to mobile agents. You could always look at your devices, your laptops, et cetera, but now to mobile devices, starting with Android and Zebra devices. You can actually see all the way to the endpoint there. Excitingly, assurance to industrial IoT networks as well as IT and OT come together. You can see it again through one topological view as if it was your own network. That helps you to measure where it matters. Secondly, seeing every part of the path. Last year, we introduced ThousandEyes for AWS. You could see inside AWS with full detailed visibility. We were the first to do that.
Secondly, this event, we're announcing ThousandEyes for Microsoft Azure as well. We're going to continue to give you insights into the public cloud providers. ThousandEyes and Splunk. In organizations, I've heard it for years. There's always the silos that exist between applications teams and network teams. What if you could actually more seamlessly integrate visibility between your network and the application itself? Let's look at an example of that. For those of you who use ThousandEyes, this will be very familiar type screen. Normally, you're using it to absolve the network. There's a problem that occurs. You're going here to understand, is it the network or is it something else? You can see that there's a test that failed there in Oregon. You click on it, you look at the path viz, and you can see that it's clean.
The network is clean. You go back there. From there, you can go directly into the APM service map. In the APM service map, look where it's red, look at the trace. You can see that there's a no method error in the application stack. That actually translates to a 500 internal error. That's what's causing the problem. We did that literally blink and you miss it. It's that fast. In the past, you've had to do human escalations from one team to another, sort of try and figure innocence across teams. With these types of integrations, it's about making your lives easier and speeding up these sort of standard things that happen in an organization around our different dependencies and the infrastructure that exists. Really excited about that. Finally, how assurance shows up in Meraki dashboard.
We've done a ton of work there. What we're providing now is an organizational wide view of your assurance health. You can look across your entire organization, drill into specific sites, look at trends, surface issues, and get into site-level details, all the way from that macro view down to a client-side view as well, where you can look at individual client health and reachability of that client throughout the rest of your network. All of this is showing up in the new Meraki dashboard. Really excited about that. I just want to say that everything we've gone through is available at the showcase behind you. It's all demoing today, right now, all this week. You can go there, have a look at it all, get your questions answered, talk about your environments. There are all experts out there.
I really would recommend that you do that. With that, I know the moment you've all been waiting for, which is to hear from real-world customers. I'd like to invite Turi to stage from Itaú Unibanco. Please join me in welcoming him onto the stage.
Good morning, guys.
Good to see you.
Thank you for inviting me.
Have a seat. You may have heard of the company. They're the largest financial institution in Latin America. With that, Turi, why don't you introduce yourself and maybe say a few things about the company?
Yeah. My name is Diego Turi IT Manager at Itaú Unibanco, leading the teams responsible for SD-WAN, LAN, Wi-Fi, and even 5G initiatives with over 20 years of experience specialized in enterprise network infrastructure. I've focused throughout my career on Digital Transformation and improving connectivity and user experience. About the I taú , Itaú Unibanco is the largest private bank in Brazil and Latin America, serving more than 70 million customers and operating in 18 countries with 96,000 employees. We require a robust IT infrastructure to ensure security, scalability, and high availability for our customers.
Amazing. Ninety-six thousand employees, that's a large number. Seventy million customers. I mean, that's incredible scale. You were saying you look after your network systems, Wi-Fi, SD-WAN. That's fantastic. Can you share actually on that point a bit more about your environment? We have a lot of networking experts here that run operations and infrastructure. Can you share more about your environment?
Yeah. Our network supports 150,000 daily Wi-Fi endpoints, including 96,000 corporate laptops in my environment. We've deployed 14,000 Wi-Fi access points across branches and offices. Connectivity is breakdown. VPN, 45% on the endpoint connected remotely. I have a more large branch is remotely in environments. Corporate, I have 22% of operating high-density areas. It's a more large office. For branch office, for branch Wi-Fi, 30% is medium-density branch office is agencies and anything. For wireless, I have 3% is an exception connection. My focus is massive resilience on wireless network. Effective management of Wi-Fi and VPN infrastructure is critical to maintain performance and availability.
That's an incredibly large deployment of Cisco devices and networks that you have. That's incredible with 150,000 endpoints and devices and a number of corporate laptops, et cetera. Let's get into perhaps, as we talk about operations, perhaps some KPIs. What are some important KPIs that you look at for Wi-Fi connectivity?
KPI is very important for Itaú Unibanco. About the MTTI, reduced man-time to identify Wi-Fi issues from 3.5 hours to 20 minutes is a 90% improvement. Using ThousandEyes and Cisco Catalyst Center for technologies for observability, Wi-Fi availability, connection success rate, average throughput per user, and employee satisfaction. I have two targets. The first is Wi-Fi health. I have 96% in this moment, but I have a target for 97%. It's a good reference for many companies in Brazil. For NPS, I have 78% for all employees. It's 96,000 employees. It's a good number. Itau is a benchmark for Wi-Fi innovation in Brazil and among the first to achieve 100% Wi-Fi infrastructure, consistently leading advancement in wireless technology.
Your use is the benchmark. And just to repeat one of those numbers there, 90% faster MTTR, mean time to resolution. That's incredible. Normally, it's that detect stage that we can see improvements in, but mean time to resolution with the integration of ThousandEyes and Catalyst Center, that's incredible. Can you give some other examples of how you're improving operations for this group here?
Yeah. I have three good examples for share.
OK.
The first is the Wi-Fi case. Before ThousandEyes, my MTTI for Wi-Fi issues was three hours due to manual analysis, issues replication, and on-site visits. With ThousandEyes AI and Cisco Catalyst Center, we dramatically reduced MTTI to approximately 20 minutes, roughly a 90% improvement. This elimination of on-site visits improved the user experience and lowered costs. About the Microsoft Teams, we use to take hours to identify root cause of Teams issues. If affecting productivity, ThousandEyes now enables real-time issue detection, reducing MTTI and MTTR from hours to mere minutes real-time, enabling faster and more precise incident response and improving overall user experience. The last case is an internet insight. Previously, external tasks were unclear for several hours, affecting operational stability. ThousandEyes enabled us to proactively detect service provider backbone issues before impacting operations.
Resolution time improved by over 90%, which incidents now identify and address nearly in real time, prevent major outages, and ensure service stability.
Fantastic. We talked about Wi-Fi, Microsoft Teams, and Internet Insights, which is another sort of ThousandEyes product. That's really material numbers that you just shared there. Can you share some maybe plans for the future? As you've heard this week at Cisco Live, and you've heard from all the keynotes, and you've met the teams, can you share some plans for the future?
Yeah. I think the Cisco Live is a rich place for new insights, for new ideas. I am excited. I have highlights for share. The integration of AI with ThousandEyes across Cisco's portfolio was impressive. SaaS monitoring and Internet Insights enhancement are really relevant, giving our reliability and cloud servers. It was inspiring to hear success stories and best practices shared by IT leaders. Excited to implement automation for combined ITSM, Catalyst Center. ThousandEyes is a pretty good tool for me, and exploring Cisco's AI solution for proactive issue prediction.
That's fantastic. Just as we wrap up here, any advice for the audience in their role in operations? Is there any advice you can give them? Maybe one piece of advice.
Yeah. For audience, I have important points for share again. The first is Cisco is a strategic partner, helping us lead digital innovation and adopt cutting-edge technologies. Invest full observability to measure user experience, not just infrastructure. We are technology leaders, and the answer to problems is really important. Measuring the experience and not just the infrastructure is the new rule of the game. Use tools capable of addressing vendor-agnostic scenarios and operational challenges. For closing my participation, thank you, Sanjay, for inviting me.
Thank you, Turi. Just appreciate your time, and thank you for your partnership.
Thank you very much.
Join me in thanking Turi. Thanks so much. Thank you so much. Just to wrap up here then, we've talked about unification, and we've talked about how we're bringing our platforms together with our unified hardware, our unified licensing, and also unified support. It's really about making it simpler to consume, to deploy, to ultimately help you in your infrastructure strategy and in your operations strategy. What's next? We're Wednesday morning, so there's a few sessions left today and tomorrow. I'd encourage you to go and attend those related to our networking advancements and operations. Also, check out the demos behind you in the digital resilience area. You'll see lots of examples of Assurance and ThousandEyes and all integrated.
The platform advantage, how you can make 1+1+ 1 = 5 or more in terms of the compounding effects, and also signing up for a free trial. Just a couple of other things. Those of you that use Webex, there's an incredible promotion this week, which is getting 50% off for Webex One, which is later this year. It is going to be here in San Diego. I'd encourage you, maybe get your cameras out, encourage you to take a snap of that and use the QR code and get that discount. Also winning this pair of earbuds. I've got a pair. It's fantastic quality. It's a really wonderful thing if you can win it. Finally, and we really appreciate this, it's evaluating the sessions, giving us your feedback.
We're always trying to make these better over time and make sure that they're relevant for you. You've given up a lot of your time to attend this event, and we want to make sure that we can deliver for that. We're not taking questions now, but we will take questions. I can be around afterwards. With that, thank you so much, everyone. Have a wonderful rest of Cisco Live. Thank you again. Take care.
Hey, guys. Welcome back. Look who's in the host chair. That's right. They let me have my own camera. My entire crew, actually. A bunch of extra cameras. Not pointed at me, because you know what that does to your weight. I'm not going to have to deal with that.
In just a moment, we're going to have that backstage conversation. Lauren's going to have it with Sanjay to go a little bit deeper around what we're seeing with Assurance, specifically AI-based assurance, where that is going, some additional details. I think that was a partner he had on stage. I didn't get to pay enough attention to what Sanjay was digging into there. The idea of AI-powered both—see, this is the problem. I can't even see my notes without my glasses. AI-powered management and insurance from edge to cloud. I think edge is interesting in terms of where we're going to see some of the newest need for network to pre-position, say, existing trained models when they go into inference mode as we get things closer to the actual need. I am looking for more information on that. Of course, we're seeing the smart switches.
We're seeing the integration of the cloud-based security, as well as Silicon One and the Hypershield with the new capabilities that we got from those acquisitions, starting to take action both from a monitoring standpoint as well as an enforcement standpoint. It is a little bit overwhelming for me, I'll be honest. I think security is doing a lot of incredible things here. Again, as I've stated earlier, assuming you've even watched everything I've done, the idea that security and AI, we're figuring this out as we're simultaneously learning about it. It is the proverbial, how do we fly this plane while also making a better plane and figuring out what kind of plane we really want it to be? Now, like I said, does that switch? Does that work at all? I don't know. Feel free to tell me.
If you guys are participating in social media, which I trust and hope you are, please use the hashtag. It's Cisco Live. So # Cisco Live. The old # CLUS has been retired. I think a lot of people are still using it, so you'll probably get some use out of it. From what I understand, we want to push # Cisco Live and keep that consistent. With that, thank gosh. Lauren is ready with Sanjay. He's just come off the stage. Lauren, I'll let you take it over to your guest there.
Hello, hello. Thank you. I am here with Sanjay Mehta. Thank you for joining us today. We heard a great session from you and had a few more questions, if you didn't mind.
Yeah, please.
The first is, we hear so much about AI Assurance, right, and how it's a nice to have. We would hear that. Why would you say it's actually a must-have, critical component?
That's a great question. Actually, this week is a big example of why it's so critical, because digital experiences are critical. It's not just human-to-human. It's device to application. It's now agent-to-agent. All of these different vectors create a lot of complexity on networking in terms of latency, in terms of throughput, in terms of bandwidth, in terms of the time to troubleshoot. Ideally, organizations want to go from sort of a reactive posture to a proactive posture. It really enables the ability to do that. What AI brings are several things. The first is, algorithmically, we can look at historical patterns and use those to predict, to see where issues could occur in the future.
Secondly, using AI agents in Canvas to just speed everything up, to supercharge our ability to actually get value out of Assurance in our organizations.
Amazing. Would you say that's--is there anything you'd like to add in terms of how that's impacting the unknown and owned assets?
Yeah, that's a great question. What we've seen, really, since the advent of the internet, is organizations now need to care not just about their enterprise networks, but with the external unknown networks. That's actually where ThousandEyes started. ThousandEyes is a technology that was really launched over a decade ago and now part of Cisco as of the last almost five years. ThousandEyes started by providing incredible visibility, the best visibility into the networks you don't own, as if it was your own network. That gave companies a real sense of control, a real sense of being able to now understand what was really a black box before. They couldn't really see into it. What we've been doing with announcements this year specifically, and it's this Cisco Live, is giving really excellent visibility into the owned networks as well.
Now you can see both the unknown and the own networks with the same level of details, with the same level of understanding. Experiences do not care. Experiences transit all networks, whether you own them or not. Yet we care, and yet we are accountable when something goes wrong. It may not be our fault, but it is our responsibility. Being able to see across all of those environments seamlessly becomes incredibly important.
I just want to pull on that thread a little bit, because it's super important to see across those environments, like you so eloquently shared. How would you say AI is transforming our precision and the speed to which we're doing that?
That's a gr eat question. There are a couple of things. I think on the front end with Generative AI, for example, you democratize access to what is generally something which more like a network engineer type person would really understand more of in terms of devices and latency and all the various KPIs and metrics. That's still incredibly important. There is a certain layer of just understanding of where is the problem that you can now democratize out, which is a lot of value. Then with AI Canvas that was shown.
That was exciting.
While the splendid.
Oh my gosh.
At the keynote yesterday, the ability to then turn that into visualizations and then persist that in that such dynamic way is going to be incredibly powerful for people. There is that. As well as visibility, there is also the ability to sort of translate between an experience and all the layers that that experience depends on. Because there is so much in the stack now. It is from a device to all the protocols, to all of the various layers to help drive optimizations of a network, all the way to the experience itself. Being able to correlate that, we use AI to do that too. Finally, closed-loop workflows. Ask any customer, they will say, yeah, seeing it is great, but can you fix it for me as well? We have been doing this for a couple of years now with, for example, Cisco SD-WAN.
We have been providing the ability to predict where there could be an issue based on historical data patterns, and then flag that, and now also automate the fix into the control plane when there is a problem. We are doing that everywhere across Cisco, because ThousandEyes, being part of Cisco, is integrated everywhere, both in terms of the sensors and our ability to actually see across all of Cisco's equipment, networking equipment, as well as surfacing insights through our own dashboards, but also Catalyst Center, Meraki dashboard, Splunk, and the new AI Canvas.
Amazing. I feel like I got to get my hands on this. And the engineer in me wants to test it. Thank you.
You can, because it's on show right behind you.
We did not even plan that.
Many places.
That was a great segue. I want to run over there. Thank you so much, Sanjay, for your time. Back to you, Robb.
Absolutely. I'm so impressed, Lauren, you with those natural setups and transitions that way you just work it and you think you don't even know what's going on. But you do. You're controlling it quite well. This is so exciting, guys. I don't know exactly what we're going to talk about next, but I do know the players involved. We've got Rupesh, who is Director, Product Manager. Product Manager, Cisco Compute. No, this is consistent. I've messed up every single title that I've been asked to use. We'll deal with it later. You've got a couple of guests with you, very important companies. I'll let you take it from here.
Thank you, Robb. Just want to introduce my two guests. I've got Will Ton from Nutanix, and I have Craig Walters from Pure Storage. Welcome, guys, to Cisco Live. So far, how has Cisco Live been? Any key highlights, any key high points that you have been seeing?
Thank you. No, I mean, I've lost count of how many Cisco Lives I've been to.
You're not going to say that.
Yeah, I've been to quite a few. I got to say the energy and the excitement for this Cisco Live has been absolutely impressive. From the keynote to all the demos that I've seen on the show floor, it's been an incredible experience.
Yeah. San Diego as well, of course. It's great being back here.
Yeah, absolutely.
Love this town.
Yeah. Let's get started. First, I want to talk about at DART Next 2025. That was just a few weeks ago in Washington, D.C. We made a big announcement about the significant collaboration between Cisco, Pure, Nutanix, focusing on a solution, what we call FlashStack with Nutanix. It's basically an integration about Nutanix cloud infrastructure software, integrating with Pure Flash Arrays and compute network environment. I want to talk about this in a little bit more detail. I've got a couple of questions. Also, for the viewers, I just want to first a little bit talk about, if you're not familiar with the Cisco Nutanix partnership and Cisco Pure. Cisco Pure partnership goes way long. It's almost like 10 years, a decade.
Over 10 years now.
10 years.
We launched together the solution, what we call as a FlashStack, which is a validated design, modern converged solution, where we integrate Pure Flash Arrays with Cisco Compute and etworking. The main reason for this partnership centers around for helping customers to deliver the validated designs and solution so that they can simplify their IT operations, but also take this design and put it in their network environment and get a predictable performance and deploy this with confidence. Recently, probably the Cisco Nutanix partnership is relatively new. It's been roughly close to two years ago, close to that. I define that partnership as a 360-degree partnership. That's a partnership where we are not only doing co-engineering work with Nutanix in developing solutions, products, and architectures, but also we have worked together in integrating our CX services with your backend support systems.
Also, with both of you guys, we are doing joint go-to-market, and our sales teams are working together to sell the solution as well. With that, I want to just talk about some of the questions that I want to kind of ask you guys and see what your thoughts are. Let's start with the big picture. The announcement we made between Nutanix, Cisco, Pure teaming up to reimagine HCI. In terms of driving this, what's this collaboration about, and how is this different from what we had before? Will Ton.
Yeah, yeah, great question. Throughout the conference, I've had the opportunity to speak with a lot of customers and partners about the solution so far. I think one common theme that often comes up is customers are really excited about having more choices when it comes to deploying or to modernize their infrastructure. With Nutanix, I think we've done a great job of offering customer choice. You can deploy any application anywhere, whether it be on-premise, in the cloud, or even hybrid. Now we're expanding on that opportunity of choice by enabling support for external storage vendors such as Pure Storage. As you mentioned, Nutanix and Cisco have had a great relationship so far, building upon our hyper-converged platform. I'm really excited to say we've now expanded that partnership with Pure Storage.
Together, all three companies are jointly developing a new solution for our customers. Another common theme that I've been hearing a lot about is the excitement around FlashStack. As a former Cisco employee and customer, I know the value that FlashStack brings through Cisco-validated designs. I don't think there's anyone more excited about that than Craig, though.
Of course. Yeah, yeah, cool. No, that's great. I mean, really, just to kind of zoom in on that a bit more, I mean, it's really important to understand what FlashStack represents. Because, I mean, again, Rupesh, you were saying we've been in partnership for over 10 years now. In that time, we've got a significant body of reference architectures, Cisco-validated designs across many different application use cases. It's great, to your point earlier about providing customers with options, optionality, and choice, and being able to support Nutanix as part of that solution. FlashStack really is just, it's not just Cisco-validated designs. Not that that's a bad thing or anything, but it's also the interoperability and the testing and the validation of the technologies that make up the stack. It's also, too, around the supportability of the solution.
Then finally, the management of that system as well. I think it would be remiss without us talking about Intersight and the value that Intersight provides a customer for that 360-degree almost management plane for the whole solution. Yeah, super excited about it.
Great. Yeah. I'm excited about the solution as well. Because, as I said, we have a validated solution with FlashStack for years. Now we are bringing Nutanix into the mix, which is what the market needs, the market requirements are. We have talked to the customers who are excited about the solution as well. I'm really excited to see this partnership going. I have a question about performance. This always comes from the customers. Performance, scale, simplicity, which is at the heart of innovation. Can you guys a little bit talk about how this new approach delivers real value to the customers on the ground? Craig?
Yeah, sure. Look, again, Pure's been in the market about 15 years now, a very similar amount of time to Nutanix being in the market as well. We started off with an old Flash solution. So we've always been Flash from the get-go. I don't know if you can actually see here, I've brought my prop here. I've actually got a direct Flash module. The whole idea with this, and I think this kind of really pulls back to what performance and capacity is about, the whole idea with this is that we have NVRAM on these DFMs so that we can coalesce writes and then acknowledge the host straight away. We destage that write then onto the Flash memory at a later time.
What that really means, and why I'm talking about this, is what it really means is it means that the application has really low latency. What that equals really at the end of the day is a really good user experience. Whatever application is running on our storage systems, you can guarantee that there'll be high performance. The other part about DFMs as well is the capacity. This particular DFM here is 37 TB, it says on here. We have a 75 TB. We have a 150 TB. Next week, at our Cisco, sorry, our Pure Accelerate conference, we will be announcing the next phase of that as well. I can't go into too much detail. My point being is that this is kind of a key factor of our differentiation from a technology perspective.
OK. What about Will?
Yeah, I think if you look at the announcement and the solution, it's easy to think that we took what we built with Cisco and then our hyper-converged platform and somehow just connected Pure Storage and made all that work. At the core of Nutanix, and I'm sure the same goes for Cisco and Pure Storage, is that our customers really value the simplicity and scale of our solutions. When we were looking, jointly developing the solution, we really looked at all areas of the product. How do we make sure we kind of continue that ethos? I remember one of the early decisions that we made, Craig, is early on when we were doing this joint development, we always knew that we wanted to keep the user management through Prism, which is our central management plane. We didn't want to deteriorate that experience.
We wanted to also make sure that we're able to leverage the performance and data efficiency on the Flash Array side. To give you an example, one of the decisions we made is that for VM operations such as snapshots or clones, rather than trying to do that within Nutanix, we're actually offloading all of that directly to the Pure Flash Array. That's just one example of how the three companies are really looking to innovate, bring something new to the market, not just do what came before us.
That's right. I'll talk about also, we just recently made M8 available, auditability on the Cisco side. I'm super excited about that. Also, we have Intersight, which we have worked with Pure, and we have worked with Nutanix as part of the solution. That just, it's a single solution that can be worked both with Cisco as well as with Pure and the Nutanix side as well. With that, I just, thank you, Will Ton. Thank you, Craig, for coming here. I'm super excited to partner with you guys and take this solution to the market. Welcome to the club and looking forward.
Let's go.
Yeah. This is great. I can't believe the amount of technology that's coming together with these, I hate the term, best of breed players. But it's like everybody's playing in their lane together. And you guys have worked on validated designs together as well. So a lot of optionality, as you mentioned, for customers, as well as just the ability to execute. And what's become an exciting environment for all of us. Thank you, guys. I really appreciate your time. Thank you as well. Lauren, we'll go back out and check with you on the show floor.
Hello, hello, hello. Yes, Robb, I couldn't agree with what you said more. I also heard simplicity. I think I heard that word simple a few times. That is what we need to drive to. Life is getting more complicated. We've heard about it in the wonderful keynote from Chuck's message. We are making it simple in our partnerships and what we're delivering to customers. Simplicity is our goal. You know what else is simple? The way for you to interact with us, whether you're here on site and you're taking a selfie at a booth. If you were in the DevNet area, hammering away on a lab, take a selfie at that computer, OK? Please share your experience with us on any of the social media platforms that you feel most comfortable with. For those here on site, remember, #CiscoLive, #CiscoLive, #CiscoLive.
If you are remote and tuning in with us, you can also participate. The community, we're very inclusive, right? It doesn't matter if you're here or not here, OK? Please hashtag us if you have a question. Do #CiscoLive. If you hear something cool, if you hear me say something cool, Robb, Steve, Z, Michelle, #CiscoLive us, @CiscoLive, mention us, right, @CiscoLive so that we can kind of interact with you and listen to your feedback. We love it. Yeah, the simple pieces is the main driving frame of everything that I'm hearing here today, especially with the last message that we just heard. I think when you think about what we're doing, try to understand that we're trying to make your lives easier. That's what Cisco wants to do. This is all about helping our customers have the best experience possible.
The first way we want to do that is simplicity. Talked about it with the hybrid mesh firewall, the simple management pane that you can use. We talked about, as we're looking at end-to-end assurance, having an AI-driven method to do it. Again, at the heart of that, simplicity. The AI Canvas, my eyes light up when I hear that. Because again, we are bringing you what multiple tools would give you in a single pane of glass. I know we overuse that sometimes, right? That is what we're doing. It's not just the goal. It's actually what we're putting into production. Super exciting, super, super exciting. You know we also are, like another thing, we talked simplicity. Another theme is AI. That's everybody's favorite, mine as well. Take a look at this video here to see more about what we're doing there.
It's a new day for the new era. AI is everywhere. So are we. We have the infrastructure AI needs. And now, the breadth of data AI craves. We'll use AI to help the world see more, do more, and we'll secure it like never before. You've all heard the AI hype. Now you want AI's help. That's exactly what we'll give you. Cisco, making AI work for you.
That was, hopefully, that was a really great video. It relaxed you. I like looking at that video myself. AI is at the center of everything that we are doing here at Cisco. I am going to pass it over to my friend, my lovely friend, my amazing friend, Robb.
Absolutely. Thank you so much, Lauren. I've got two very special guests with me. I'm struggling a little bit because I haven't played on our partner program in a while. I used to work in channels heavily dependent on CDW. I'm glad to have you here as well. I can't tell you how many times you've failed me and my customers that I was responsible for at one point in time in many different types of situations. Old news now. Let me have you introduce yourselves. I have consistently messed this up all day long. I don't want to do you the injustice. Please tell us your full name. And what are you responsible for?
Yeah, absolutely. My name is Elisabeth De Dobbeleer . As you can hear from my accent, I'm from Belgium. I am a Senior Vice President in the Global Partner Sales Organization. I am in charge of the Cisco Partner Program, which is now undergoing a major transformation. We're on a journey to the Cisco 360 Partner Program.
I'm going to ask you more about that in just a moment. First, let's find out, Karl, what is your full name? What are you responsible for there at CDW?
Karl Schulz. My title is Vice President, Product and Partner Management at CDW. My team and I handle and manage the overall relationship with Cisco from a strategy and business plan perspective.
I'm always impressed with just how deep the rabbit hole goes when it comes to our partner programs. Because I've worked in a lot of different areas, profitability is always a top concern for how do we leverage this expertise that we have. You guys are making changes. We're looking to get even better at this. You mentioned, is it called 360 or Partner 360?
The Cisco 360 Partner Program.
360. That's why they don't give me official names. Tell us a little bit more about what this is and what the timeline is.
Yeah, absolutely. Last year, at Partner Summit, we announced that we are undergoing a huge program transformation. Why that is, is that we are basically overhauling the way we define partner value, as in the value of our partners and our ecosystem brings to the market and to our customers in this whole world of AI and fast developing technology. The program consists of a new way of measuring partner value along four dimensions. The first one is what we call foundational, which is all about the partner's ability to engage in a lifecycle practice and managed services practice. From there, we build up to a specific investment per portfolio, because it is per the six portfolios, networking and security, et cetera. Their investment in technical capability, whether it is things like CCIEs, et cetera, but also our Black Belt training.
From there, it's all about our partners contributing to our customers loving and using product, which is onboarding, adopting, et cetera, which makes it easier to renew. That's what we call engagement, again, per portfolio. Finally, it's about what we call performance. Because ultimately, we want to work with our partners on growing together.
Yeah, we've got to move some stuff. Yeah, move some services and things. I'm going to just assume that CDW is a good partner for us. Or you would not have invited them here. probably should not cut you out of that part of the conversation, Karl. No, these are so, well, reflection. What does this mean for CDW? How are you reacting to this? What are you?
You know, it's certainly very impactful. And we anticipate that as the program details continue to roll out, there's a lot here. We're, number one, we're really trying to prepare. We're really sitting down and doing the analysis to understand some of these anticipated outcomes. What we've also been doing is sitting down with our Cisco counterparts to really get like a deep understanding of those partner value indices and the foundational capabilities that those are calling for. Just trying to map that against our current capabilities, our certifications, our technical specialist organization, some of our enablement. We're feeling really good so far about how we rank up to those expectations and where we're going to land. Part of this is also about, as we think about, where is Cisco signaling for all this to go? That's the part that kind of gets us excited.
It seems as though what Cisco is saying is the customer outcome is really at the center of all of it. I think that for us, it aligns really nicely with how we go to market anyway. Less of a singular architecture focus, perhaps, than maybe in the past, and more of this multi-domain approach surrounded by services, surrounded by some best-in-class CDW capabilities. To us, it just seems like this natural extension of how we already go to market, and certainly with Cisco. No unnatural motions that we're seeing so far. We're getting excited about it.
CDW, I think, can sometimes I know I've made the mistake in the past of short-shrifting the notion of what you provide. Because you do have a full suite of services. I would assume you've probably been looking at lifecycle for quite some time. These are not new to you. Aligning with Cisco from your relationship with Cisco is one thing. Aligning with your relationships with Cisco sellers and those relationships that you've had for a long period of time is not always simple. It's a process. It sounds like we're beginning that process. I'll ask you about timeline in a moment. How do you feel about alignment with Cisco sellers?
Number one, I have to say, for us, it's always best in class in our partner ecosystem. So we're very appreciative of it. We look at the program, and we sort of think that the structure and the framework could maybe even enhance it further. The Cisco and CDW sellers, as they sit down with a customer and they ask that customer, what are the issues you're trying to solve for? How do we help drive some outcomes that are going to grow your business? There's a strategy that's very meaningful and important to that. It seems as though the Cisco 360 program now has this sort of check-in and natural built-in collaboration to get both sides to focus on every step of that lifecycle as we go past the initial transaction, adoption, expansion, and then in a long-term success.
If we're more effective working together, who's going to benefit? Well, ultimately, the customer. The customer is going to get the support they need for the full lifecycle. And then we're all going to be sitting on this tremendous amount of success.
Yeah, well, that's what we're going to pull off, right? What is the timeline, Elisabeth? And what kind of things should other partners collectively look at for preparing? Because it sounds like there's a little bit of work involved, a little bit of awareness, I'm sure, is also something you probably.
Yeah, there's obviously a lot of work involved. Because we are talking about transforming a program that is, in some elements, more than 20 years old. Like VIP, for example, which is one of the crown jewels, let's say, of the existing partner program, is celebrating its 22nd anniversary, I think, this year. It is quite a task. I think one thing I want to say is that I really want to thank Karl and all the partners out there. Because when we announced that Partner Summit last year, we said three things. We said we're going to take time, 15 months. It is implemented or launched in February 2026. We still have seven months to go. We also said we're going to do it together. The level of engagement and leaning in of CDW and also other partners is just, it's been really amazing.
I mean, I was just commenting to Karl. We have this Webex space with our top 75 partners. They are constantly pinging 24 by 7 all around the world. There is an ongoing engagement and dialogue. Because we really want to get this right. The third piece was protecting the investment. This is then touching predictable profitability, which is obviously a number one priority for our partners and a number one priority for us as well. We really want our value proposition in this new program to be very compelling so that our partners really want to choose to work with us. That is really important.
We are very committed to consistent profitability as we make this transition and making sure that our partners know how to shift in their ability to earn and to shift with us as we go through the product innovation, the CX innovation, and the sales motions evolution as well.
Does this spell the end for any other profitability programs that we've had o r do those kind of continue?
They're being collapsed.
They're collapsing into it as well?
Yeah.
Because some of them have been very successful, I feel like, in terms of just figuring out how to safely raise the flag and say, hey, we want you to tell us that you're working on this so that we can provide resources appropriately, not to shift a deal in any certain way or anything like that. Also make sure that we do not misrepresent partner capabilities and that we engage them early and often, which becomes important, I think, to preserving your profitability.
To seller alignment. Yeah. One thing I want to say as well, that the things that are working well, like again, the VIP logic of you book and you get a certain predictable percentage on that, that logic is continuing in the new integrated Cisco partner incentive as well. We're not breaking the things that are working.
Each of you, just real quick, we've got about one minute left. Things you want people to remember, and please make it CDW specific. What would you like? What do you wish more people would understand about CDW in the Cisco audience?
I'd love for the audience to just walk away understanding that CDW and Cisco couldn't be more aligned on how we approach not just the market, but customers solving for outcomes with the full customer lifecycle. I would love for everyone to understand that CDW has best-in-class capabilities in that regard, but also that we are extremely appreciative of the partnership, not just with Elisabeth, but with the entire Cisco team. Really happy to be here.
Elisabeth, I'm only at 10 seconds. What do you have?
It's all about growth. It's all about benefiting from the amazing innovation and what's going on in the world right now.
I'm going to have to toss over so we don't miss anything on center stage. Guys, we'll check back with you on the other side. Thank you so much for joining us.
First being AI-ready data centers, second being the future-proof workplace, and the third is digital resilience. This talk is going to focus on the AI-ready data center and how we can infuse security into the fabric of our network in a way that I think is fairly, fairly unique. You never want to discount how significant the change that's happening all around us really is. If you had a chance to listen to the keynote, we had Kevin from OpenAI talking about these kind of rapid advancements in the models themselves. That's really the engine behind all of this.
We expect to see very significant adoption of new kinds of applications that are going to call for a new approach to networking, a new approach to compute, and a new approach to security. That is why we created this organization at Cisco, is that we are looking at bringing those three things together in a way that I think can solve problems in a fairly unique fashion. Our experience is very much informed by experience that we have working with some of the largest hyperscale AI factories in the world. These are the folks that are building the models. They operate at scale that is really unimaginable. You have probably read about this in the paper. They are literally buying nuclear power plants. What could possibly go wrong? They are buying their own nuclear power plants to fuel these things. It is that level of scale.
We're working on models that have a trillion parameters. In this process, we've learned a lot about how do we optimize the network to deal with this type of compute. Because when these customers spend literally billions of dollars on GPUs, you need to make sure that you're feeding those GPUs in an efficient fashion. What we have found is that the network is actually the bottleneck here. When a GPU is talking to another GPU, they do it in a way that is synchronous. If there's a little tiny glitch in the network, you lose the whole job, and you have to restart the whole job. When we think about the networking fabric, we tend to divide it into two buckets. We talk about what's called a front-end network. A front-end network is the network that we're all very familiar with.
It feeds your web servers, your database, your traditional VM-based applications. The difference between an AI-based application and a traditional VM or Kubernetes-based application is that the traditional apps that we've been working on for decades have all been based on CPUs, which are sequential processors. Each time that Moore's law clicks, the processors would get like twice as fast every two years, so double in performance. With a GPU, it's a parallel processor. Each time Moore's law clicks, not only does the processing element get faster, you get more processing elements. What this means is the amount of data that a GPU can process is like a leap more data than we've ever had to deal with before. This creates a really interesting scale argument. We are currently shipping state-of-the-art silicon that can run 800 Gbps per port.
Now, if you're working on a 25-gig network trying to transition to a 100-gig network, you're like, I don't think I need 800 gigs. You will when you get these AI workloads. Because it's the GPUs that are talking to the other GPUs that are burning this much bandwidth. What we're finding is that 800 gigs is not enough. These GPUs are still saturating the network. The trick to a back-end network is to take Ethernet networking and get it to scale up to deliver the very, very high bandwidth that those GPUs need across an individual cluster. Also, we want to go across multiple clusters. There is a whole bunch of very specific capabilities that we're going to talk about that are built into your Cisco networking.
Because our philosophy at Cisco is that whether it's a front-end network or a back-end network, it's all network. One set of management, one set of devices, one set of tools for troubleshooting that can distinguish between a front-end and a back-end network. Now, we're able to deliver on this because we have a unique advantage in the industry in that we're the only networking vendor that has the full stack. We have the silicon that we can design ourselves. We have the optics. We have the systems and, of course, the operating system that we put on top of it. We have another advantage in that we have a deep and very meaningful partnership with NVIDIA. We are the only vendor that has an NVIDIA and Cisco validated Ethernet design for these back-end networks. That's super, super important.
Because, like I said, you want to treat the network as the network, whether it's the front-end or the back-end. Now, I realize many of the folks in the audience are looking at this and be like, oh my god, these giant GPU factories, that seems I'm not going to buy a nuclear power. I hope you're not going to buy a nuclear power plant. We do not want you to. We do expect that customers are all going to be experimenting with AI because it's moving so fast. It really is so transformative. Like, let's just talk about software development. I run a very large software development organization. When I look at the tools that my team is using, tools like Windsurf and Cursor, these tools are as effective as a software developer.
We're seeing a 30% increase in productivity, a 40% increase in productivity with these tools. That's not limited to software development. This is happening in every different function, call centers. Throughout the whole business, you're going to see these AI-based applications, which means you're going to get your requests for getting AI infrastructure up and running in your environment. Devil's in the details here. A 96 GPU cluster can have 1,000 cables, 1,000 network cables that have to be put together, lots of little ports, a lot of complexity, very, very sensitive to the implementation. These things can't miss a beat. One of the things that we've done is we're bringing a level of automation to the whole lifecycle that I think is very, very innovative. It's particularly relevant for these AI back-end networks. It's what we call Hyperfabric.
We're going to talk more. Murali is going to come up, and we're going to talk about Hyperfabric and how it can simplify the mundane job of cabling the 900 or 1,000 cables you need for these GPU clusters. We want to make this stuff work for the enterprise, not just for these super-duper AI factories. We're going to talk in a little more detail. Murali, why don't you come up and join? Murali runs networking, data center networking here at Cisco.
Thank you, Tom.
Thanks, Murali.
Thanks, everybody. Thanks for taking the time.
I think Murali got bigger applause than I did. Oh, it's because he's wearing a coat.
I planned to be the audience.
Yeah, yeah, yeah. It's all his friends that are here. Cool.
Yeah, exactly.
Smart move. OK. I think I set the stage. AI networking is different than just front-end networking. It has special requirements. And we're seeing this in real life. Yes?
Absolutely.
Yeah. When we think about this NVIDIA design, we've been working carefully with NVIDIA to make sure that both the front-end network that you know, but the back-end network that ties these GPUs together has unique capabilities. Maybe you could touch on some of the things that we're doing there with NVIDIA to deliver these validated designs.
Absolutely. One of the things that we've noticed is that in these environments, as you said, data-intensive flows, we have synchronization not just within the GPU, but across the network, being able to parallelize and then reorder and reassemble on the other side of the network. The third thing is we were able to build out congestion control in the network. To do that, what we need is to be able to address a highly low-latency, ultra-low-latency fast environment. We need to be able to provide things like intelligent packet flow. This intelligent packet flow basically is what has been built as a secret sauce for us with NVIDIA to allow for that spraying of packets.
Yeah. These back-end AI networks, it's almost like TCP doesn't work there. Because they're so latency sensitive. It's not a web server.
Yes.
We have to compensate for that and build mechanisms that are going to allow these things to work.
Yeah. Your traditional ECMP is not going to work here. What we had to do was we had to provide new techniques, new load balancing techniques, like flow-let load balancing. Also, we wanted to add the adaptive routing or the packet spraying capability that NVIDIA and Cisco jointly synchronized on. That allowed us to build this bucket of capabilities called intelligent packet flow.
Yeah. Now, the important thing here is that these are unique capabilities that we call AI networking. Guess how you use them? It's part of the Nexus switching infrastructure.
Absolutely. It absolutely starts from the differentiated silicon, the systems that we build around the differentiated silicon, in this case, Silicon One, and also the software NXOS that we've written, and the operating model, the Nexus Dashboard that we will talk about a bit more.
Yeah. In networking, management is always the hardest part.
Yes.
Having insight into the workload that you're providing the network services for is, I think, a fundamental capability. Talk about what we're doing to be able to track the AI jobs.
Absolutely. We talked about Nexus Dashboard. We've seen the keynote earlier from Jeetu and others about the unified Nexus Dashboard. That applies to AI environments as well. We want to be able to monitor networks, whether that's front-end and back-end. Like Tom just said, we want to look at managing of networks not as individual silos, but across front-end, back-end, management, storage environments. With this, what you see is you have a site-level view of your AI wherever your networks are being deployed. You click zoom in to the particular network. In this particular case, what we're going to be doing is going into the back-end network and looking at the various jobs that are running there. Those could be also jobs in terms of latency, but also anomalies within the jobs that might be existing.
We go and zoom in into the, particularly in this case, the first job. You see that there are some anomalies there. Having that visibility in near real time is important. Being able to do that and then pinpointing that this was because of a specific traffic degradation due to a particular GPU-to-link pathway. That particular degradation, getting to that very quickly, and then ultimately narrowing down to the point that it was because of excess temperature thresholds being reached, that ability to get that in near real time is very powerful.
Yeah. Cisco has always focused on bringing the network closer to the application. This is another instantiation of that where we can understand and interpret the AI workload cooperating with NVIDIA. We get that information from their management systems to help you pinpoint the problems and figure out, is this a network problem? Is this a GPU problem? And get the job back up and running.
Absolutely. Absolutely. This is quite critical to correlate and triangulate all of those together, nothing more than the unified Nexus Dashboard as well.
Yeah. Let's talk more about that unified Nexus Dashboard. That's been a big body of work for us, right?
Yes. One of the key things that we've seen is we have 40,000 customers operating, as you said, different kinds of environments. Front-end has a mix of CPU, GPU. Back-end has complete GPU-based environments. You have storage. You have SAN environments, a lot of environments. Ability to manage them in different disparate silos is no longer going to work. That is why a unified Nexus Dashboard bringing together various architectures across data plane, control plane, policy plane was critical. That is what we've achieved.
Yeah. People have been waiting for this for a while. It's here, right?
Absolutely. Absolutely. People have been clamoring for this. What we've done is not just have we provided this across the Nexus Dashboard, the data center side of the house. We've expanded this onto SAN storage, as well as your third-party environments as well. This is open interoperable framework fabrics. You can actually use this across a number of environments, a number of services, including media, environment fabrics, data broker environments, all kinds of environments.
EVPN from the data center all the way to the wireless, right?
Absolutely.
Yeah.
One thing I want to point out is, as Tom just said, EVPN across all of these environments, it means that when we brought all of those architectural enhancements, you also have to be able to define centralized policies and be able to enforce them across these environments and segment them. We'll talk more about segmentation with the next section around security as well. This is integrated into that as well.
Yeah. Yeah. One of the things Cisco has always done well is not only build the networking devices, but thinking about how do you use this stuff. We have introduced new operating models over time. If you remember, we had original Ethernet networking. With ACI, it was really a new way of operating the network. We've introduced a new innovation here that we call Nexus Hyperfabric. This is an even higher-level abstraction that makes these complex network deployments much simpler to turn on, to operationalize, and to troubleshoot over time. Let's walk through an actual example of what Hyperfabric can do. It all starts with a cloud-based interface. You log in. Simple, modern, easy. It's going to present you with a dashboard that shows your existing fabrics. Now I would say, I want to add a new fabric.
Remember that 96 GPU cluster we talked about with 1,000 cables? Let's go through what steps it would actually take to design and deploy this. You click a button, say, I want to add a fabric. You get a bunch of templates, small, medium, large. Let's pick the medium one. You're going to go through, and you can edit that template. You can customize it to exactly what your deployment is going to look like. Here's your compute. Here's the networking. Now notice all the little details, the racks, the bolts, the cabling, all of that stuff is all done for you. That used to be on a spreadsheet, right? It was kind of a long spreadsheet. Then you lose the spreadsheet or you forget about it. Right?
A lot of operational challenges with that.
A lot of operational challenges that we're using AI back-end and cloud management to just abstract all of that out. Now watch. You push a button and say, yes. It will automatically generate that bill of material. It automatically placed the order. A bunch of boxes show up on your shipping dock. You open these boxes. These systems are pre-configured. You've got a wiring diagram where it's going to tell you step by step. Remember the 1,000 cables you've got to plug in? There's a little iPhone app. It'll show you, plug this cable into this port, plug this one into that port. It tests every single time you make one of those connections, it tested you do it right. You can actually mess with it and be like, plug it directly into the wrong port.
It'll detect and say, hey, wrong port, buddy. Move that connection and put it in the right location. Simple, simple, simple. This does not stop with fabricating the network. This works when you're troubleshooting operations as well. You can go in and take a live system. You can pull a cable out. It'll detect that. It'll route the traffic around it. It'll alert you, hey, you just lost a cable here. It'll tell you where to go find that cable.
Absolutely.
This is material, right?
This is material. In fact, I was just at a customer, enterprise customer, who said that I'd like to get my back-end environments running up and running as quickly as possible. An environment like Hyperfabric for AI will allow you to deploy that immediately. More importantly, once you hit these kind of issues around cabling, or you pull a plug, or some of those things that humans do, maybe that's going to go away with physical AI, but humans do in the near term. What you'll have is you'll have immediately the ability to see where exactly. In this particular case, what you're seeing is that there is an exceed threshold that's showing up in a storage to the network part connectivity. That is very quickly, simple, easy UX. Be able to see this across the environment. You can click on the details.
You can see which particular pRobbes have detected higher latency. You can see that there are certain flows, and there are certain segment percentage of them that are actually at a higher latency. The rest are in line.
Yes. We're making it reasonable to take one of these networks, these back-end networks, and deploy them. It can be very large scale, right?
Absolutely. Massive scale networks. Essentially, if you look at these clusters, even one AI cluster could actually look like it has more bandwidth than many of the telcos combined, right?
It's an amazing statistic. One rack of AI compute has more bandwidth than the entire mobile networks of North America. That's the world we're living in. This is today, not like five years from now. That's happening today. What we're giving you is an operating model that will deliver the performance, the capability, but also the management so that you can actually run this stuff in your infrastructure without crazy extraordinary effort and crazy extraordinary amounts of power and space and cabling, et cetera. Simple is the name of the game. Let's talk about the Nexus Dashboard and where we're going here. We've been able to take all of that integrated management, and we also draw telemetry from that and integrate it into Splunk and ThousandEyes for observability. We're going to talk a little bit about troubleshooting.
Absolutely. One of the things we've focused on is providing insights from the fabric. Obviously, the network sees everything. The network can span across the data center, the connectivity side, the WAN, as well as into the campus environments. Being able to, from the data center, being able to stream out the data, collect them across all kinds of architectures, like we said, with unified Nexus Dashboard, and then push that out into Splunk. You have connectors that connect into your Splunk cloud environments, or for that matter, other environments on-prem as well.
One of the things I think is most exciting about the combination of Cisco and Splunk together is we're looking at ways to provide insight into that networking data that's generally not observed because it's much too big. You can't ingest NetFlow into a typical platform like Splunk because it's just too costly. We believe there are ways to ingest that information, the relevant information, to provide the insights necessary without driving a huge ingest bill. You'll be hearing a lot more from us in that domain. This is the first step, delivering that integration. Thanks so much. Now we also want to think about some of the devices that are powering this revolution. One of the things that I think we've announced recently, but it's worth mentioning, is this category of device that we call a smart switch.
We have two flavors of smart switch. We have 9300s. We have a spine configuration and a leaf configuration. What a smart switch does is it really brings two personalities together into a single device. It's got a packet forwarding engine based on Silicon One that moves those packets fast, efficiently, and reliably. Then we add security processing that's executed on a DPU, Data Processing Unit. We partnered deeply with AMD to deliver that in a single chassis.
In fact, it's right there.
Oh, yeah, yeah. There we go. Yeah, yeah. Right? These big ones are the Silicon One. These smaller chips here are the DPUs. This is going to transform how we do security because it allows us to take what used to be a firewall and break it literally into a million pieces. Imagine that? A million little tiny baby firewalls where you can put individual security under each piece of your application. We have an AI management capability that will make that all possible. This architecture is more than just a feature. I think this is really a new class of device that brings security into the fabric of the network. High performance, AI ready networking with the ability to integrate security into the fabric. This is something I think is really kind of unique to Cisco.
There are others that talk about it, but this is really in our wheelhouse. We have a unique capability here. That is not all. Optics are a critical part of this. As we move into these very high performance networks, you're going to be seeing more and more optics. One of the things we've announced at the show is a new 400 Gbps bidirectional optic that can take advantage of existing optical cables.
In fact, Cisco was the inventor of BiDi as a concept itself. Taking that all the way from not just from 25 Gbps to 100 Gbps, 100 Gbps- 400 Gbps is amazing. Being able to provide that at scale is really nice.
Operational efficiency, simplicity, but performance and scales for the AI era.
Reuse all your cables.
There we go.
Yeah.
Murali, thanks very much.
Thank you.
Now, the other side of the coin here is we need the compute that's going to drive all this. I'd like to invite my friend Jeremy Foster to come up. Let's talk about what we're doing in compute and how we merge these things together.
Absolutely.
Jeremy runs all of our UCS and Compute business. And Jeremy, you've been busy, right?
Been really busy. I think, in fact, the compute business until G2's kickoff yesterday was probably one of the best kept secrets at Cisco. I mean, for the folks that have the paddles, since we have not used them yet, how many people knew we have almost tripled the size of our UCS portfolio offerings? Everything that you see here today is actually shipping. All right. I have got a few yeses, but a lot of nos. That is good.
Those paddles are pretty big. I hope we don't have a paddle fight going to break out here. That'd be a problem.
The important thing here is this. It's choice. We have a full offering of not just modular servers or blade servers where we lead the industry, but a full offering around AMD, Intel, everything you would need inside that rack mount portfolio. We are extending that, of course, into AI, where you can go over to our booth and check out the new C845, which is a scalable GPU server that allows enterprise customers to start off with two GPUs, four GPUs, eight GPUs, and scale as you go, which is really important because a lot of enterprise customers, Tom, to your point, just getting started.
Just getting started, right?
You don't want to overspend.
GPUs are not inexpensive, right?
Yes, exactly.
Scale as you go is important. Now, everything we talk about, management is the hard part here, right? And management means not just how we manage our stuff, but how we fit into the other management tools in your environment. Talk about our ecosystem.
Yeah. You've got to have the right building blocks and piece parts. The ecosystem in the data center is one of the reasons I love working in this space because we can't do it all ourselves, right? All of you as customers have a ton of things that you've integrated your operations with. Whatever those things are in your tool chain, whether it's ServiceNow, for example, where both Nexus and UCS have great plug-ins for you to be able to leverage to make your lives easier. We want to take all the partners that you see on the screen, whether it's the ServiceNow type folks or the storage partners, the Pure Storage, the VAST Data of the world, so we can build end-to-end solutions and take a lot of that guesswork out that you have to do when you move into this AI era.
You have to figure out not only how to build a front-end network, a back-end network, different types of storage that also require networks. Putting all that together can be complex and a little bit of a daunting task.
Now, where we really shine is the integration around these AI workloads. Let's talk about AI pods and how do people experience that.
AI pods, we talked about those last year when we first got here. We launched this product to focus on inferencing use cases in the enterprise because that's where a lot of folks were starting. What we announced this week was expanding those, as well as the server portfolio, to cover that full spectrum of things that you would want to touch, from training to fine-tuning to inferencing. We can bring that all together in an AI pod experience. What is an AI pod? We take that storage, our networking from Nexus, our compute from UCS, the software from NVIDIA. NVIDIA's AI enterprise software really brings it all together to make it very easy for you to execute your use cases. We put it all together for you.
We can give you a Cisco-validated design, which means you can pick up the phone and call Cisco to get support across that end-to-end stack.
Yeah. That includes the interaction with NVIDIA, right? I think this is a really unique capability.
Tom, I brought you a little gift.
Whoa. He bought me a Nexus series box. This is pretty small.
It is small, but how many people like Legos? Our booth's right over here. You can leave here. You can go build your very own AI pod and walk away with something to take home. When you get home, you can call your account team and you can buy one for real, too.
Now, management, management, management. This is where Cisco has some real unique capabilities.
Definitely.
Talk about Intersight.
This is where we lead the way with Intersight and UCS overall is management. It's all about making things easy. We've continued to not only invest in this portfolio and making it so simple to roll out a server, whether you're doing that in the data center for a traditional use case or for an AI use case, and in the future even for things like edge use cases. We're adding new capabilities, including next-generation metrics. What that means is I'm sure you've all used the widget on your phone, and you can kind of set that up in a dashboard. We can track metrics over time, things like power, cooling, CPU utilization.
For example, you can make yourself a widget that says, "Tell me all my servers that are using less than 10% of their CPU cycles over this period of time." You can then look at those and know what servers you need to go shut down so that you can save some power, for example. New advanced topology views. Some of the things you saw in that Hyperfabric demo where you had those really pretty topologies that show you how everything is connected, we can do that in Intersight as well. It's really slick looking. The complete visibility insurance just means now that you have this great topology view, we've given you a lot of ways to look at and understand the performance of the system end-to-end from compute through the network and through the storage.
Now, when we think about the full stack experience, the network and the network services are really the most important part of the computer. It's what ties that whole infrastructure together. Cisco's taken an important step forward in this regard. We want to look holistically. We want to be able to deliver all of what you need to get a workload up and running with an integrated approach. One of the more important parts of this is load balancing. Jeremy, pop quiz for you. What is the most desired feature from a load balancer in the eyes of a customer?
At the risk of probably missing something, balancing loads.
Yeah, balancing loads. Exactly, right? With ISO Valant, we have a very, very high performance software scale-out load balancer. The idea here is this is so automated, it does not have a million knobs and levers and dials. You're not writing I rules and custom tuning this and that. It balances loads. It is software that can run at the speed that approaches hardware load balancing. It is heavily optimized using eBPF, which is a capability that allows us to make kernel optimizations and process the packets as efficiently as possible. This is widely deployed in hyperscale and cloud environments. You have big websites that are using it. It is deeply integrated into Kubernetes.
As we start to think about this data center stack of the future, this ISO Valant load balancer, which is available today and customers are using, we have enterprise customers that are beginning the adoption of this. We think this is going to be an increasingly interesting part of this fully integrated data center stack.
Tom, it's great, too, because not only did it start at the big hyperscalers and it massively adopted, but we're making it easy through the validated designs to bring it into the enterprise. We're also bringing it into that ecosystem with folks like Nutanix, who has Nutanix Container Platform that are already integrated with Isovalent. You can take advantage of it there as well.
You're alluding to a very, very important fact, which is it's our view that Kubernetes is the platform of the future. Kubernetes is the platform that will run not only these AI workloads, not only container-based workloads, but also VM-based workloads. I don't know if you all know that, but Kubernetes is actually quite capable of running a VM-based workload. I spent five years working at VMware. And all of my customers are asking me, "Tom, where do we go?" The answer is Kubernetes. Now, how do we get you there? This is where things get a little bit complicated. That same eBPF and Cilium capability that I talked about that has the load balancing that we acquired with the Isovalent acquisition, that same capability can provide very unique VMware to Kubernetes migration.
The challenge we see is that what VMware does unbelievably well is that when a VM comes up, it can talk to all of its peers without actually knowing anything about the physical infrastructure underneath. This is what enables vMotion. You can take a live VM, move it from one machine to another, and all the traffic just automatically goes with it. As you think about this migration, you want to be able to move a VM one at a time into a Kubernetes environment. Kube can run those VMs. The challenge is a networking problem in that Kubernetes is built for the cloud. It does not know anything about Layer 2. That is not a notion in Kubernetes. What we have built is a bridge where we can connect between the Layer 2 networking constructs of VMware and the Layer 3 constructs of Kubernetes.
This allows you to move the VM without changing the IP address of the VM, which is a really big deal. Each VM can move one at a time, and it can reach back and talk to all its friends. Migration is not something you're going to do 10,000 VMs at a time. You're going to do it one at a time and allow those VMs to seamlessly connect back to the rest of the cluster running in the vSphere environment. You can slowly make your migration work. What's important about this tool is that it is independent of the underlying infrastructure. This is why Kubernetes is so important. You can make a migration to the public cloud. You can move from the public cloud to another public cloud. You can move it back to a private cloud.
You can run on OpenShift. You can run on a different flavor of Kubernetes. You can run on Nutanix, right? We have the ability to provide you the flexibility to optimize the workload for the infrastructure that's available to you. That infrastructure might change over time.
Yeah.
Now, this is the heart of what we think is this new data center architecture. All of the fancy AI networking that we talked about, the load balancing that we talked about, was kind of oriented around these AI-based workloads. Our thought is that same stack, in the fullness of time, will mature so that you can have a single Kubernetes-based management with all the network and network services being implemented in things like a Cisco Smart Switch, sitting on top of a UCS box. That's your new data center. This is what we're delivering, is that management layer that's built around Kubernetes and the infrastructure that runs underneath it that can run a front-end network and a back-end network, a Kubernetes workload and a VM workload, an AI workload. It's all managed from the same platform. This is our opportunity.
I mean, that's the key point. You can't set up a new silo every time you have to deploy a different type of technology. I think when you look at this diagram, we're also calling out, yes, we're going to bring things together in Cisco Cloud Control. We're also going to make sure that the folks that are operating Nexus Dashboard, Intersight today, those element managers don't go away.
Correct.
We're just going to bring it all together and make it better.
They get sucked into this larger platform. Jeremy, thanks so much, my friend.
Pleasure.
OK.
Thank you.
The last thing I'm going to talk on is some details about how we actually bake security into the fabric of the network and how we can really change the way we think about things like a firewall and how a firewall operates. It starts with a secure foundation, physical root of trust, post-quantum crypto built in and DDoS built into Cisco networking equipment. Now, here's where we've done something really innovative. The industry is talking about hybrid mesh firewall. The idea behind a hybrid mesh firewall is it allows you to have multiple different form factors of enforcement, but a single policy layer.
That single policy layer is done by Cisco Cloud Control, our cloud-delivered management that can push your rules into a hardware firewall, into one of these eBPF clients that I talked about from Isovalent, into one of these magical DPU Smart Switches, and also into third-party firewalls. What we announced at the show is two new advancements to our hardware appliance firewall appliances. We have a low-end firewall, which is extremely competitive price performance, the 200 series. It runs 1.5 Gbps of firewall and IPS. We have a high-end firewall, which can run 400 Gbps of firewall plus IPS. That is very, very high density. It is 200 Gbps per rack unit. It is a very efficient form factor. That really rounds out our product line. We have refreshed all of our firewalls.
We think we're very, very competitive in this space, including virtual firewalls. Where we shine is not only do we have price-performance advantage with our hardware appliances, we have the ability to deploy this firewall in what could be hundreds or maybe even thousands of locations, right? Hundreds of thousands of locations, little tiny miniature versions of the firewall that sit right next to the workload wherever the workload is running, all managed from this single interface. In this firewall architecture, we have a deep understanding of the workload. As the workload starts to evolve into AI and they start to have things like models and they start to have things like logic streams that need reasoning, we have a new set of capability that can understand and interpret the behavior of your model.
We have the ability to both validate and protect your AI workloads and apply what's almost human-like reasoning to say, is this thing behaving the way I think it should? This can all be done as a feature on this distributed firewall. Now, we've seen some pretty significant shifts in the industry. There's been a series of attacks last year called Volt Typhoon and Salt Typhoon, which were targeting critical infrastructure: telecommunications, water supplies, power equipment generation. They were targeting the gear that runs in that critical infrastructure. That's mostly Cisco gear. We were very, very involved in these types of attacks. We take them very, very seriously. Our approach to this is not like, hey, we're going to try harder and make sure there's no bugs.
We have a new approach to how we can address security, which is about the ability to address what we call the patching hard, the patching gap. Patching is hard. Patching is complex. Patching an application is hard. Patching a switch or a router is hard. This is why many times customers will be running a version that is two years old. With a compensating control, we can identify the vulnerabilities that are running in your application. Let's say it's a Postgres database and say, oh, in this distributed hybrid mesh firewall, we recognize that we're running under a Postgres server. We're going to apply a compensating control that can shield the vulnerabilities in that Postgres server. We can do this in a very automated fashion. Now, importantly, a compensating control does not eliminate the need to patch.
You think of it as a finger in the dike. It's a shield that blocks the vulnerability. A very smart attacker can find a way to get around that shield and still exploit the vulnerability. It allows you to patch in an orderly fashion. Patching isn't going to be an emergency anymore. The system is designed to be highly automated. It applies the compensated control when you go about and patch. We remove the compensated control so you don't have some giant collection of firewall rules that nobody really knows what they do. This is all implemented in that hybrid mesh firewall. The hybrid mesh firewall is a mesh, which means it can live simultaneously in the operating system using eBPF. It can live in one of Jeremy's UCS servers that has a DPU, one of these chips.
We also have networking devices like the one we're showing here, where we have DPUs that are running in the switch. What this means is that every switch port can be a high-performance Layer 7 firewall to apply those compensating controls to the application. Now, if you're listening carefully, you'd be like, hey, that's cool. What are you doing, Cisco, to protect the network operating system that's running on the switch itself? Here's where we announced something I think is extremely innovative. We introduced Cisco Live Protect. Live Protect gives you the ability to apply a compensating control to a vulnerability in the network operating system. You can do it without rebooting the switch. It's a live compensating control that can be applied to the switch, which means if we find a vulnerability, we will present you with this compensating control.
You manage this from the Nexus Dashboard that Murali was talking about. Apply the compensating control to the switch without rebooting the switch. It is going to transform network operations. I don't want to ever be in a situation where I have to call you and say, you know what? We've got a P0 vulnerability. You have to take that system offline and patch it immediately. That is very, very disruptive. This solves that problem, I think, in a real unique and innovative way. OK, we talked about AI workloads, what they're going to look like. They're changing the nature of compute. They're changing the networking requirement to backend them. We're pulling all that together, integrated solution. We talked about how we're bringing security and infusing it into the fabric of the network with things like smart switches.
The last piece of the puzzle is the connectivity that ties all of this together with our service provider partners. To go through what we're doing in the service provider space, I would like to invite my colleague, Guru, up to the stage. Guru.
Hello, Tom.
Good to see you.
Hi, everyone.
Once again, Guru gets more applause than I do. Murali gets more applause.
I got the memo.
It's the jacket. It must be what it is. OK, Guru, service providers have been in business for a long time. All of a sudden, we have this AI revolution kind of changing everything. What does it mean to the service provider?
Yeah, so everything we talked about so far has been about how do you produce AI, whether it's training workloads, inferencing workloads, hosting of AI workloads. All of what we talked about was the infrastructure to do all of that. What we are starting to see now is the adoption, the using of AI. As that grows more and more, all of that AI traffic, most of it is actually going to go over service providers' networks. This is what service providers see coming. You know those three names that you showed, AT&T, T-Mobile, Verizon, they're going to see a whole lot more traffic as more and more people start using AI applications. Service providers know this. Their fundamental priorities haven't changed. They still need to provide secure, resilient connectivity as efficiently as they can.
They also need to try and find ways to monetize and get more revenue-generating services. Now they have to do it in the era of AI, right? That is what they are looking at.
We have an answer for this, right?
We do. We looked at all of the pieces that need to come together for service providers to evolve their networks for the next 10 years. We built an architecture that we refer to as the Agile Services Networking Architecture. We call it an architecture because it's not one thing. It's all of those pieces that need to come together. What are those pieces? First and foremost is the fundamental building block of a network, which is a router. We've innovated in routing, introduced an entirely new portfolio called the Cisco 8000 for service providers. The routers need to be connected. We have a new model of connecting the routers together by converging IP and optical layers. The routers need to be, that network needs to be managed. We have automation. Now we've infused AI in the automation.
Some of the stuff you saw in the keynotes, that's all part of our service provider management portfolio as well. Finally, security. That needs to be infused into the network. Tom referred to some of that. All of those pieces coming together in a validated design is what we have brought together.
Yeah. So let's talk a little about the specifics here. What are we doing in routing? You mentioned that that's a big core focus area for us.
Yeah, so let's look at specifically what is the innovation we are introducing in routing. I mentioned that we have a new portfolio for our service provider customers. This is based on our Cisco Silicon One technology. Here are two examples. We introduced this portfolio three months back. Now we continue to add to it. These are the recent additions. One of them is a new access router. We call it a converged access router because it can converge all different kinds of connectivity endpoints, whether it's broadband, whether it's satellite, you'll hear more about that, or regular Ethernet and IP. All of that can be converged into a single device to provide that connectivity piece.
We have a new series of edge routers, extremely dense, 6.4 TB of capacity that can do full service provider features and scale in essentially a small box, one RU box. This is exciting stuff. It allows us to completely reimagine how the networks are built and really make them a lot more efficient.
Just like we talked about in AI networking, Silicon One powers these solutions. Optics are a critical part of the overall solution, right?
Indeed. That goes to the connectivity piece. If you think about how connectivity has historically been done, it's a layer of IP routers connecting to a completely separate layer of optical routers. What we have done is taken a lot of that optical transponder gear, and we can put that in these small pluggable form factors that you see here, put them right into the router. We can eliminate an entire layer of optical gear, which makes for a far more cost-efficient router. We call this routed optical networking. We have new pluggable optics for all kinds of distances, right? Thousands of kilometers, 400 Gbps, 800 Gbps, 100 Gbps, as our customers may need for connectivity.
Just like in the data center, management is the name of the game here, right?
Management has become extremely critical. What we have done is the same AI technologies that you saw demoed in the keynote, AI Assistant, for example, that is now coming into our Crosswork portfolio. It allows us to do things like troubleshooting, proactive management of configurations, and all driven by AI. It gets a whole lot more efficient. There is a lot more detail to unpack here. I would highly encourage you to double-click into this. This is just a flavor of what we are doing right now.
These things you're talking about are not theoretical, right? These things are deployed with some customers. Give us some examples of what it's doing.
Everything I've talked about, except those two platforms that are coming in the next three to four months, is here already. Many of our service provider customers have actually already adopted this. They're seeing huge benefits. You can see some of the names here. This is coming from our customers themselves. When they put the Agile Services Networking Architecture and implemented it in their networks, they're seeing efficiency gains like 60%, 70% OpEx savings, CapEx savings, and power. Those are not numbers to be ignored. We're not talking 5% or 10% incremental gains here. This happens because our platforms are fundamentally more efficient. We can deliver a lot more bandwidth at the same scale in smaller form factor devices instead of large chassis. We can connect them and eliminate entire layers of connectivity. That's a lot of power, space, and cost savings. It's huge.
You can take a look at some of what our customers are saying here around it.
Guru, when I was at Cisco in 2007, so many, many years ago, John Chambers stood up and he said, I see a world where we're going to put routers in space. Yeah. And I thought, I don't know. Right? I don't know. And he wasn't wrong, right? It's actually happening.
It's reality now.
Yeah.
One of the things that has happened in the Satellite Networking Technology Space in the last few years, if you think about 15 years ago, 2007 time frame, Satellite Technology was geostationary satellites, way higher up in space. You could not get enough bandwidth. The latency sucked. What has happened now.
Speed of light is just a fundamental limitation.
Exactly. You could not overcome laws of physics, right? Now with LEO satellites, low Earth orbit satellite, everybody—Starlink started this, but everybody is doing it in a big way. Everybody is launching entire constellations of satellite much closer to Earth, right? Much lower in space. That has made satellite connectivity really high bandwidth, low latency. It has become a very viable connectivity mechanism. What we have done is partnered with Starlink and others and integrated our terrestrial networking, all of the portfolio I talked about with Agile Services Networking, with the likes of Starlink. Now anybody, any service provider that says, I need to provide coverage in certain areas that I cannot lay fiber to, for example, they can look at a validated design from us, buy the Starlink equipment, buy Cisco equipment, put that in the ground, and it is just going to work.
What's more, historically, service providers had to manage these two networks completely separately, right? You couldn't get visibility into the packets going through space and back to the ground station because it was a completely different environment. We've integrated that telemetry too. A service provider can get complete end-to-end visibility across the terrestrial and non-terrestrial network. It's really exciting stuff.
A router in space is tough to upgrade, just to say. It actually works, right? I think it's going to be a really fascinating development we're going to see in the very near term. Guru, thanks so much.
Thank you, Tom.
OK, to pull it all together, we've got three customers. I'd like to ask our customers to come up. Michael Israel from the Kraft Group, Cameron from Groq, and Shawn from Dakota State University. Come on up. Hey, Shawn. Please sit down. Cameron and Michael, good to see you. Michael, good to see you again. Please have a seat, gentlemen. All right. I have got my Johnny Carson or David Letterman style questions here. I am going to throw them away. I will kind of work down the panel. We will start with Michael. Can you maybe actually, each one of you could just give us a quick introduction on what you do, and then we will talk about some kind of specific things. Starting with you, Michael.
Good afternoon, everyone. I'm Michael Israel, Chief Information Officer for the Kraft Group. I run IT operations for the enterprise of Kraft operating companies, Gillette Stadium, New England Patriots, New England Revolution. We also have several paper-based businesses. We have a recycling plant in Connecticut, 10 cardboard box manufacturing plants on the East Coast. We have a commodities business that moves products all over the world, paper-based products.
Can you talk about how the data center modernization has helped you keep up with NFL speed?
Sure. My job is very interesting. Everyone only wants to talk about the NFL. Not paper.
What about the paper? Yeah, yeah.
We implemented Cisco Fabric now just about two years ago. It's running all of our AV operations at Gillette Stadium. We were one of the first stadiums to bring everything into an IP world from production control. Anything broadcast, AV related, all of our audio, video, IPTV.
That used to run on these very expensive dedicated networks, right?
Correct.
Yeah.
We took all that proprietary out and brought in a Cisco Network, which has been running, like I said, for the past two years. It has been a really great experience with Cisco Professional Services leading the way.
We see a similar thing happening where some of these AI deployments require one of these dedicated proprietary networks that we're moving towards replacing with Ethernet. That's what we talked about earlier. Can you talk about what AI is going to, how it's going to impact your operation, the teams, and the NFL? Where do you see AI actually being used in your environment?
When it comes to the NFL itself, we're at the infancy right now. Because anything you see on field is actually controlled by the NFL proper. When we look at player training, prepping, practice, that type of stuff, we're just now starting to touch on, gee, how can we help augment what the coaches are doing, training players? We got into video-based anomaly detection now, again, a little over two years ago, in which we can train models on what's happening in the stadium or in our manufacturing processes. We're looking at, typically, we have, as an example, 550 cameras, security cameras at the stadium. We are reusing that data and teaching AI models to detect everything from overflowing trash cans to God forbid there's a fight or an assault in the stadium. In real time, acting on these things as they happen.
Yeah. It truly is transformative. Cameron, you're on the cutting edge of this. Cameron at Groq, you guys have done one of the fastest. You built a very, very large-scale data center in a time frame that I didn't even think was possible. I think that probably made for some late nights. Can you talk about some of the critical infrastructure decisions that were wrapped up in that?
Yeah, certainly. I look after Network and Data Center Engineering at Groq. Groq is very fast, extremely fast inferencing capacity. To do that, we need to build very fast. For Cisco, really where Cisco have enabled us is to accelerate that speed of those builds. Looking to Cisco for switching, optics, and getting that supply chain assurity to be able to deliver into our data centers very quickly is where we see the most value.
Those 800 Gbps super-duper switches and the optics, this is what this guy's actually using. Can you describe the business impact that that's had?
Certainly. The transition from 100 Gbps to 800 Gbps has been transformational for us from a patching perspective inside the data center, where we can deploy DR8 optics, which is the 8 Gbps by 100 Gbps. One thing that's been very transformational with the Cisco Silicon One has been the extremely high RADX and the ability to get 512 100 Gbps ports from the Nexus platform.
Yeah. You guys, I mean, I'm just so impressed with the work that you're doing. The app just keeps getting stronger and faster and better. Looking forward to more. Now, Shawn, when we think about higher ed, you guys are just starting on the AI journey. Talk to me about the challenges that a university has in managing infrastructure.
Yeah. We have a diverse campus. We have research space. We have our students, our academia, obviously. We have production. We're expanding out beyond that in contractual services as well. Formerly, we operated somewhat independently and siloed. That's a challenge for us. We're always asked to do more with less. For us, we really had to look at our technical strategy and to provide the additional resources that we need. We needed to be able to borrow resources from the central repository. We can no longer operate that way. We need to have better performance. We need to have visibility. Cybersecurity in everything we do. We are a cyber university. It's always a priority for us.
The ability to scale is augmented with what we're doing, centralization of data center services in a type of hyperscaler model.
Yeah. You have everything from students who are ordering pizza to sort of researchers that are working on proprietary data and stuff in between, training grounds, et cetera. Your infrastructure needs to be flexible enough to deal with that, right?
Yes. Flexible and agile. Absolutely.
Yeah, yeah, yeah. What led you to work with Cisco to modernize that infrastructure?
That is a long story, kind of. I started with Cisco back in the late 1990s and just had a lot of exposure. I managed network infrastructure for a railroad. We built large networks there. I am a champion of Cisco just because I have really not worked with other companies that have the full stack technology solutions that Cisco has. You cannot find a company that can also provide what Cisco can. For us, building our requirements, we just lean on Cisco to help us to architect and to deliver those solutions that our businesses need.
Yeah. Yeah. For our viewers out there, I sort of thought maybe this is a beer. It's not. Yeah, it's just water. Didn't it kind of look like it could have been a beverage? No, it's water. Damn it. OK. For each of you, you've each been through transformation, albeit totally different context. Maybe we'll go down. I'll start with Michael and work down the line. Talk to us about what our listeners, I mean, there's a very, very large online group that are going to be listening. What should IT leaders be thinking about as you go into this transformation in the AI era? Michael, I'll start with you.
It's a rather interesting time in which our infrastructure teams, our security teams are having to catch up with the business. We're usually leading the charge. With all of these AI use cases coming forward, our business and our stakeholders are coming out saying, hey, can I do this? Can I do this? What about this? What about this? We're looking at not just can we deliver in a timely fashion, but what's the risk of bringing all this AI in? What's the hidden costs? What's the compute costs? What's the cost every time we send something up to the cloud and someone hits return? More importantly, looking at our data governance, looking at our security posture, and then being in touch with what our stakeholders are actually trying to achieve.
We are really having to dive into our business units to understand not just how they do business today, but where they want to evolve to.
Yeah. The risk assessment is something I think is really challenging because AI-based applications behave completely differently than a traditional application. Here is a simple way to think about it. When a model learns something, it never forgets. You cannot go in and delete the data from the model. You heard yesterday, Kevin from OpenAI was talking about this. The more data the model has, the more effective the model is, right? The model is going to learn all your secrets. How do you control this? We are still getting our arms around that as an industry. It is a radical change. It is happening in months, not years. Now, Cameron, you guys are making a lot of these waves.
Yeah. I think certainly my sort of pitch to enterprise is maybe think about data repatriation from the cloud and where the next generation of AI compute is going to exist and ensure that you have access to your data and you're able to process it, where that next generation of AI inference will be.
Yeah. Do you have an opinion on that?
Yeah. It's certainly interesting that for the training data, we certainly see large bulk access. For inference, it needs real-time access to that data and the request to be able to be processed quite quickly.
The compute's going to sit near the data.
Yeah.
Data gravity. Yeah, yeah, yeah. OK. Shawn.
You know, for us, it's important. These aren't technology initiatives for us. These are business initiatives and business strategy. It's important for us to align with our stakeholders. We have a campus of cyber geniuses, right? We lean on them to help drive our initiatives. Because of that, it's caused us to have to modernize our data center. We really needed to do more and to have the ability to research with AI, to utilize it, to have that AI-ready data center. For us, we really let the growth of our organization, our research in AI and our research in quantum to really predicate the direction that we go.
Awesome. Gentlemen, thanks so much for joining. Thanks to all of you remote as well as here in the studio. OK? Thanks.
Thank you.
Take your business into the new era of Customer Experience with Webex AI Agent, which answers questions and fulfills customer needs, all while delivering personalized conversational experiences. With Webex AI Agent, you can engage your customers through the channels they prefer, using natural language and human inflections. Sounds like a special occasion. OK. Based on your description and our stock, I would recommend daffodils. Oh, I want a dozen, please. Understood. A dozen. Coming right up. OK. I think we have what you're asking for. I've sent it directly to your phone. Even if your customer changes their mind mid-conversation.
I've got it. Miami. That's the destination. Of course. I've found several flights that fit your preference.
With reliable guardrails in place, the conversation stays within the AI Agent's capabilities.
Could you also book me a taxi, by chance, for when I land?
I'm sorry. Unfortunately, that is outside of my scope. Anything else I can help you with?
You can be sure that your customer's needs will be met, their problems resolved, and that all-important transaction completed.
Let's get that payment arranged. I've sent a text message with the confirmation. Have a great day.
Welcome to the new era of Customer Experience with Webex AI Agent.
Thank you for calling. You're caller number 25. Your estimated wait time is 37 minutes.
Welcome back to the Cisco TV studio. It is so good to have you all with us here on the live stream. We do appreciate it. Thank you for staying involved and engaged as we move from the opening segment this morning into the keynote and into these great center stage sessions and these great deep dives. So much phenomenal information coming to us, really more than we can handle even here in the TV studio. We are having incredible interviews, great guests that are walking in the door, the number of releases, new products that we are hearing about, incredible. You are getting a firsthand approach to all of it. I am Steve Multer. Great to have you with us. We just came out of another phenomenal keynote deep dive powering the AI-ready data center. This is an enormous topic across the entire show.
AI readiness, not only in the data center, but also across the network, being able to power AI capabilities within our organizations because it's an extraordinary bandwidth demand. So many capabilities. This time, talking secure, scalable infrastructure. How do we not only secure the network to keep it from being an obstacle toward progress, but how do we accelerate innovation based on the security that allows every one of our users to access all of their applications, to utilize any device, to have proper ZTNA in the infrastructure and in the foundation of what we do? As we just heard in that keynote, it's an unprecedented era of innovation. AI, cloud, computing, IoT, digital streaming, they talked about, all redefining what's possible, all pushing the network toward greater bandwidth, greater scalability, more agility, and all based on Cisco's AI-ready infrastructure. We've heard about it again and again.
I'm here in the studio right now. We're going to go wide because Z is hanging out with me. Hello, my friend.
Hello. Hello, Steve.
First chance we've had to play together today.
No, today I've been missing you.
They keep us away from each other because they know the blend is just nothing but a challenge, right?
They're haters. Haters.
No, they're not. They're the best people in the world. You know what? Do me a favor. Would you give the social media a hit to everybody as a reminder to stay in connection with us?
Yes, absolutely. You know what? Because even though you're not here, you still can engage in the experience. We would love for you to react, connect with us, #CiscoLive2025. Send us your reactions, your comments, whatever you want to put in there. We want to stay connected because we have individuals that are waiting to consume what you're saying, getting it back to the right people so that we can answer and deliver on what you're looking for. I tell you, Steve, you said a mouthful. You took the words right out of my mouth. You know, when they were talking about the infrastructure, I thought about, you know, if you don't have a good foundation, you know what? Whatever you're building, it's not going to be stable. You are talking about security. You are talking about scalability.
Because face it, innovation means nothing if you are not secure and that businesses aren't ready to do what they need to do to advance and make the impact with their customers.
That is exactly the right point, Z. It is so strong. Again, we talk about it again and again. You can build the greatest widget in the world, but if you do not know how to make that widget applicable, if you do not make it secure for people to access, if you do not put zero trust in place so that the right person is accessing the right capability, the widget dies. This is such an important thing to bring up again and again and again. It is hard to believe we are already more than halfway through the conference. We are well over halfway through the conference at this point. We have seen so much excitement here on the show floor, and we have seen so much engagement. I had an opportunity to talk with Oliver Tuszik a little bit earlier, and we are going to replay that.
He kept going back again and again to the foundation of not only taking our technology and putting it into the hands of our users, making their lives easier, going for that better outcome, that Customer Experience, but also talking about how we make AI applicable to what is going on in their organizations. We hear again and again, you have to use AI. AI is important. Go get AI. Deploy AI. People are saying, well, how? What am I supposed to do?
Yeah.
Partnership with Cisco, right?
Partnership. That's what it's all about because it has to be practical. You can't just be in theory. They need our help. We're here to give them what they need to be trusted because they can trust us, Steve. I mean, we're Cisco.
We're Cisco.
That's why we do what we do.
We didn't even take it further than that. We are Cisco. I'll tell you what we're going to do right now. We're going to head on out to our friend Lauren, who's out in the show hall with Eric Knipp, Senior VP, Cisco Customer Experience for Americas. Hello, Lauren.
Hello. Hello. Yes, I am here with Eric Knipp, Senior Vice President of Cisco Customer Experience in the Americas here. Eric, we've known each other a long time. I'm honored to be here with him today. Eric, please, can you tell us a little more about the findings that we've heard in Cisco? I have to read this because it's a really great report name. I don't want to mess up. The race to an agentic experience, how Agentic AI will transform Customer Experience.
First, Lauren, it's great to see you, and great to be with you today. It's awesome to be here on the show floor at Cisco Live. We are on the cusp of something amazing. Agentic AI is going to change the B2B customer relationship. Customers aren't only ready for it. They're actively expecting it. 56% of the customers that we surveyed said they expect it to change their interactions just within the next 12 months. That number skyrockets to 68% if we think about over the next three years, right? Again, just think about our own lives. We're used to that already. This is coming in business very, very quickly. It isn't just about the efficiency gained. Customers think that it is going to, or those that we surveyed, think it's going to fundamentally transform that experience.
92% of the customers that we surveyed said that complexity is becoming a problem. 93% said through Agentic AI, through changing that experience, we can deliver more personalized, more productive, and more predictive services. This isn't only about making this more efficient. It's about a better class of service. Customers not only want it, they're demanding it. We cannot remove the human element here. This isn't about changing or removing humans from the conversation. It's about making them better. The respondents that we asked who said, or that we talked to, 89% said that human empathy, again, the human in the element, we've talked so much about that this week, having them as part of that closed-loop system is going to be critical. We can't underscore the importance of ethical AI. Again, so important, right?
Again, 99%, almost 100%, so I want to know that 1% that wasn't there, feel that this is going to be really important as we continue to look at governance and we continue to look at how we're making sure that human in the loop and that AI are working in concert together to make sure that we're doing this with a responsible approach and ultimately driving that next generation of experience. I mean, Lauren, it's the exciting time to be a part of this.
It's a really exciting time. I'm glad I'm here at Cisco to witness it happening.
We're doing it together. We're doing it with our customers. We're doing it with our partners. We're changing the world.
Thank you, Eric.
Lauren, it's so nice to be with you.
It was so nice to be here with you. That just piques my interest. I hope all of the viewers out there also feel the same way. Thank you, Eric.
My pleasure.
I'm going to go to my next guest. We have another lovely guest over here, Jay Kuhn, Principal Architect in our Product Management Team. Jay, can you talk to us more about what we're doing with digital resilience for the data center here?
Absolutely. I'd love to tell you about a service we have. It's digital resilience for AI infrastructure. I want to show you based on an example, OK? Consider a large insurance provider who's getting in the business of wildfire detection using drone footage. Here's how this system is going to work. Drones from different geographical locations are going to send their video footage into an AI-ready data center. Think about an AI pod. On this data center, a vision language model is actually going to be processing that footage. For example, in the top left corner, do you see fire? Answer yes or no. The model is processing this footage ongoing, and it says no, but it will quickly change to yes when it sees fire. All right.
In terms of digital resilience, for this system to be reliable and resilient, we need to be able to anticipate disruptions. The question that we would want to ask here is, is the model processing footage properly? Is there any performance delay in this actual processing? This is where digital resilience for AI infrastructure comes in. We're going to take customers' business goals, and we're going to translate those into tangible steps to achieve resilience. Based on the use case, we're going to develop golden signals that we're going to implement and baseline. We're going to bring a personalized, unified view of correlated insights, bringing data from across the stack as well as business insights. Our delivery experts are engaged, familiar with the use case to provide situational analysis and recommendations. All right, let's see this in action.
The vision language model and the golden signal in this case is end-to-end request latency. This is the time it takes for the model to process footage. Right now, it's less than one second, which is really good. We can also flip over to the infrastructure and actually see the GPU utilization. It's less than 70%. This is also very good. Let's move forward in time. For this customer, business is booming, and more and more drone footage has been coming in every week. Nobody called ahead to the infrastructure team to give them a heads up about this increased demand. Look at the situation now. There's fire, but the model says no. What's happening? It's not model hallucination. This is actually a lag in model performance, OK?
It's not obvious, but this is happening to all the footage, whether or not we see fire at the moment. This is no longer just a performance issue, OK? This is a risk to the business, a risk to revenue, a risk to potential to reputation, and even lives, OK? Thanks to the golden signal, we already know that this is happening. The end-to-end request latency on this model has creeped up to 31 seconds. That's a 30x increase, OK? There's a lot of metrics we could get. This is an observability cloud, by the way. A lot of metrics we could look at to characterize this, but it's actually GPU performance. The GPU is fully utilized, and so that's causing a lag in performance here.
Thirty seconds may not seem like a lot when it comes to fire detection, but 30 seconds could turn into 30 minutes or longer if more drone footage were to come in right now.
Amazing, amazing. Thank you so much for that, Jay.
Thank you.
We are going to toss it back over to the studio. I know it's time for us to give it back to Steve.
I appreciate it. Thanks so much, Lauren, and our appreciation to Jay and to Eric as well for doing such a great job and walking us through that. You know, one of the big highlights here at the show as people are walking around in the showcase down in the world of solutions is the NOC and the SOC. They are right next to each other. This is where you get to see firsthand what it takes to deploy, to manage, to secure the Cisco Live network. Same thing with the SOC. Behind the scenes, look at active threats. We are going to go to a video right now where Robb had an opportunity to tour the NOC and the SOC. Let's send it out to that video tape.
Hey, guys. Robb Boyd here. They told me that I could share with you some of the new stuff and stuff that I think is interesting with the NOC and the SOC. That's the Security Operations Center. Let's take a look. I'm always fascinated by how many people come together in advance of a program like Cisco Live. How Cisco says, you know what? We've got to do our technology ourselves. Of course, there's a team to do that. There's a lot of young people that are learning how to do it better. There's a huge community that comes together, both Cisco and partners. You're used to seeing kind of the big, pretty wall of stuff, right, that says things are going on. It's actually real-time information coming through on here. One of the first things I've noticed is that we now have Splunk dashboards.
We had Grafana and other type of open-source type ways of doing this. Now, of course, we have Splunk in-house doing wonderful things with them that brings up security, which I promised you. I'll get back to that in just a moment. If you come by here, this changes constantly. It's always hard to say exactly what you're going to see. You get the full network layout so you can understand how all the gear is configured and, of course, how it's performing as well. One commonplace that people come to is the actual gear itself. I just want to mention that they are doing these Q&A sessions. Very interactive, very transparent. There's multiple opportunities for you to be able to talk to the NOC engineers about the actual deployment.
Because these guys deploy conservatively in the way I like to think of it. They're looking for solid, advanced technology but not bleeding edge. We may show you some bleeding edge stuff that has yet to be deployed or recently deployed out here on the show floor. These guys are dealing with the same things that most of us deal with back in our normal day jobs, which is how do we actually get stuff done safely, securely, with low risk, high output, least few people as possible. You see all the standard gear in here. Minor changes throughout. Long-time relationship with NetApp and FlexPod. That is going strong and fantastic. They've got some additional security components in here. They've also got some AI going on, not specifically laid out, but some of the AI that they're using in various ways.
I don't have the details on exactly what it is. Maybe a good time to come in and ask them questions. That's just an idea. You're obviously going to come by here and check out the equipment. Now, what's different, having followed me over here, Kevin, is you notice we have this new fishbowl. This is not NOC staff. I thought this was Network Operations Center people at first. It's not. It's Security Operations. Security obviously has always been a part of the show. Many of you have experienced different things that have to happen from a security perspective. Now we actually have full-time security analysts who are analyzing all the traffic from here, not just keeping it to themselves, but actually sharing back with people here.
We've had customers privately and one-on-one identify to say, hey, your email's in the clear, or your password's being passed in the clear. Did you know? Here's how you could fix it. How can we help? Also, of course, vendors are setting up a lot of infrastructure here to be able to do the demos. They're driving those demos from some cloud-based platform elsewhere. Security is of utmost importance. There is the wireless install, which I think was just updated with all new Cisco gear in the weeks leading up to this. We're all on nice, comfortable wireless LAN controllers and the latest from Cisco's wireless. If you get by here, you can come by here and knock on the cage. Don't feed them. There's a reason why they're in their own room. It's security analysts.
They have a little bit of a different beat. Very smart people watching out for us. You've got a few dashboards that you can reference over here for understanding what they're doing in real time. The last thing I want to point out, because I always find this fascinating, there's yellow cables up here. They're kind of hard to see. You've got two pairs of fiber optic cables. The entire show is going through those cables into the live Network Operations Center being analyzed by the Security Operations folks. A lot of great stuff here. I hope you get a chance to check it out. We'll have more information and detail as we talk to the experts that set all this up. Stay tuned.
Fantastic job, Robb. As always, thanks so much for bringing the story of the NOC and SOC directly to us. We are going to head back out to the world of solutions and center stage. We're going to learn how Cisco XDR is changing the game today for Security Operations. You'll hear from a great customer, Steve Madden, how they deployed accelerated threat response with XDR. Enjoy it. We'll see you back here on the other side.
Hello, everybody. It is great to be back at Cisco Live here in San Diego. It's nice to see some familiar faces in the crowd. For those of you I have not met, my name is AJ Shipley. I'm a Vice President of Product for our Threat Detection and Response Portfolio at Cisco. Show of hands, how many people have heard or are familiar with Cisco Extended Detection and Response? OK, only a few. Also a show of hands. We'll kind of keep this interactive. How many people in the audience would consider themselves Security people versus maybe Networking people? Security people, show of hands. Networking people, show of hands. Cool. I hope that the session today is very informative for both of you guys. What we're going to do is we're going to cover three things.
I'm going to share with you a little bit about what Cisco Extended Detection and Response is and really how we're very excited about what we're doing with this product, especially in the realm of Agentic AI and the AI that we're embedding in the product. The second thing I'm going to do, maybe a little bit different from other folks who've done these presentations today, is we're going to spend some time doing a live product demo. For those of you who've ever done live product demos, that can go one of two ways. We'll find out how it goes. We think we've got a hardwired connection here. I'm going to walk you through some of these new capabilities that we announced just a month ago at the RSA conference. We're going to finish up.
I'm going to invite up a CISO from a very strategic partner of ours at Steve Madden stores, Kamran Siddique. He's going to talk to all of you about how he's been able to operationalize Cisco Extended Detection and Response along with all of his Meraki devices, 3,000+ Meraki devices, to get the full value out of his investment in the Cisco Network. Sound good? That's what we'll cover. For those of you who have not been following along, it was just two years ago at the RSA conference, which is the world's largest security-specific conference. It's held every year in San Francisco, that we announced Cisco Extended Detection and Response.
What this product is for is specifically to be able to detect when a bad guy or a nation-state adversary has compromised an organization, gotten into their network, and either accessed some data that they shouldn't have accessed, exploited vulnerabilities, maybe stole some intellectual property. We launched this product two years ago, arguably maybe a little bit late to market compared to some other vendors who are out there, like Palo Alto Networks and CrowdStrike. We really focus on three core differentiators in order to differentiate our product from what some of the other competitors in the market had. The first differentiator with XDR at the time, two years ago, was we said, we are going to integrate with other third-party vendors out there.
For those of you who've been following Cisco for any period of time, you know that Cisco would integrate with any product out there as long as it was a Cisco product, right? We recognized that in the security space, it was unreasonable to expect a customer to go all in on the Cisco Security stack when they had made investments in a CrowdStrike endpoint or Palo Alto firewall. We said, if we're going to be successful here, we have to recognize that at the end of the day, the competitors are the adversary. They're not the other vendors in the market. We need to integrate with them. The second thing that we did was we said, we need to build a product for the masses.
Unfortunately, for far too long, the security industry has catered to the largest 500 organizations in the world, the Fortune 500, the big banks of the world that had 200 security operations people who could operate these products in order to be able to detect and respond. The reality is that every organization in the world is dealing with things like ransomware and crypto mining. Security vendors were not serving that segment of the market. We wanted to build a product that that segment of the market could get value out of so that they could protect themselves from all of the things that were coming at them, even if they did not have two or 300 people in order to be able to do it.
The third thing, for you networking people, I think this is where it's going to become really relevant, is we said, we really have to play to Cisco's strengths. Bad things happen in networks, not when things get exploited out on the edge of those networks. Bad things happen when bad guys start moving laterally through the network to get to the high-value assets in the data center. Who better than Cisco to be able to see that lateral movement occur across the network to be able to detect and respond when bad things are happening in the environment? That third thing, we really pivoted hard and integrated all of the Cisco networking capabilities to be able to differentiate how we can detect and respond to bad guys when they've gotten into an environment. That was two years ago.
Since then, it's been a very busy seven quarters, candidly. We just passed $140 million in total bookings. We just closed our biggest deal ever three weeks ago. For all of you partners out there, a $10 million 90,000-person organization in Germany that had multiple Meraki sites just purchased Cisco XDR in order to secure their environment. Next year, we have even bigger ambitions. Four weeks ago or so, at the RSA conference, we announced XDR 2.0. We really focused on three new core capabilities that all leverage this new form of AI, specifically Agentic AI, to be able to automate the thing that takes the most amount of time in an organization when they're trying to detect and respond, which is the investigation process. We call that instant attack verification.
That instant attack verification is driving a whole new user experience to make it possible for security professionals and non-security professionals, which we're going to test here in a second for all of you in the room who didn't raise your hand when I asked if you were a security person, to be able to understand what is happening with an incident and be able to describe it to somebody else. Because if we can do that for all of the people who haven't spent their time in security, then we can open up all of the available talent for these organizations so that they can go and get the people to work in their environment to be able to detect and respond, as opposed to have to battle it out for a bunch of limited resources.
Then the third thing we introduced, which I will not demo today, but actually if you walk over to the show floor and see XDR being demoed, is this new forensics capability, which is for the first time, I think, from a Cisco perspective, light years ahead of the best of what any other competitor can offer, whether it is a CrowdStrike or Palo Alto Networks. With that, I am going to flip over to a demo. OK, I am going to start here from Meraki. How many Meraki customers or partners or people who work with Meraki are in the room? OK, what you will see here is actually me starting from Meraki. If I go into Meraki and I come over here to the Security Center, the first big kind of capability that we introduced with XDR is we natively integrated it into the Meraki dashboard.
Because if you think about the types of customers that have Meraki, they tend to be maybe smaller customers, right? They might have a leaner staff. They might not have a whole bunch of IT professionals. Oftentimes, the people who are responsible for the network are also responsible for security operations. What they wanted is they wanted a place where they could actually go manage the network and manage security from a single pane of glass, if you will. With XDR being embedded into the Security Center of Meraki, now I'm an analyst. I come in today, and I can immediately see all of the incidents that Meraki picked up that XDR detected, and I can start to go work them. I'm going to change my filter here. I'm going to scroll down. Now I see one that's assigned to me, AS. I'm AJ Shipley.
This was assigned to me. If I was to just click on this incident, what you'll see here is a pop-out of some of the things that Meraki and that XDR integrated with Meraki started to see for this incident. You get a short description. This was all generated by AI. You get a list of assets that are kind of showing up. There was a device, and there's a user, Alex Wilber, that was involved, and a bunch of observables. Now, if I'm a Meraki admin, I can assign this to somebody else. I could close it out. If I really want to drill in and understand a little bit more, I can come in here, and I can view this incident in XDR. Now from the Meraki portal, it immediately takes me into the XDR portal.
Now, I'm not going to go do a full-blown demo of XDR right now, because what I want to do is I want to show you all of those new capabilities that we introduced. This incident that I saw in there involved a couple of IP addresses. There were a number of processes. This is a capability that has been really well received for the first seven quarters in XDR. We took 34 different events in this case, or observables, and we tied them together and put them together in a way that started to describe an attack. What we recognized is that there was an opportunity to do a lot better here.
What I'm going to do after I show you this next set of capabilities is I'm going to ask you, do you feel like you could follow along and describe to somebody what happened with this incident? Just hold that. You're not going to hurt my feelings if you say no. That's going to be my test for you, OK? With this new incident alpha, the goal really was to drive understanding in under 30 seconds or less. If I pop into the new incident alpha, ignore the attack graph on the right-hand side of the screen for a second. What I will tell you, though, is that this is the exact same representation of information as the last screen that you saw, that more simplistic one with some straight edges. This is the exact same representation. We will come back to this.
The really cool thing that we announced four weeks ago at RSA with Cisco XDR is the use of multiple agents, Agentic AI, to go off and start investigating all of the things that happen in an incident. What does the threat intelligence tell me? What do my firewall logs tell me? What are my Meraki MX devices telling me? Then put that all together in a summary that explains the incident and gives recommendations on what to happen, what to do next. The first thing you'll see is that this overall incident has a criticality or a high critical nature with an 85% pRobbability. How do we know that it's 85% critical? We'll come back to that in a second. It gives you a very quick description. It highlights some critical security issues with a device.
In this case, the device is named Mozzarella, some remote access Trojan activity, and some suspected data loss. You get a summary here. Again, this was no person involved in this. This is all Agentic AI that is generating this. A summary of on March 27, 2025, the XDR network identified some potentially malicious activity. You can scroll down. Cisco Secure Endpoint detected some malicious scripts that were running. There were further findings involving command line encoding, excessively long network connections. Again, you get a pretty good description. Not only do we have an 85% pRobbability that this is a critical incident, you also want to ask ourselves, what's the impact to our organization? Is this a high, medium, or low impact? In this case, four gigabytes of data left the environment.
You don't have to be a security person to know that when f4 Gbps of data leaves your environment and it goes somewhere that it was not authorized to go, there's probably an issue there. If those of you who've been doing this long enough can remember the Sony hack in 2014 or 2015, a bunch of movies left Sony, went over to North Korea, presumably, right? This is an indication of some type of similar thing happening. You also get a set of recommendations. The AI not only can tell you what happened, it will give you recommendations on what actions to take next. For example, you might want to isolate a compromised device, an endpoint device. You might want to block some malicious IPs.
Using the SOAR capability that's built in, you'll have the ability, or your users will have the ability to take this action right from here. Now, this is the really important part. How many people fully trust AI right now? I don't. How many people have a Tesla? Nobody has a Tesla in the crowd? One person? Two people? Do you allow the Tesla to drive for you? You do. You've tried it. Do you let it drive you home? Yeah. With your kids in the car? No. OK. We don't fully trust the AI right now, right? Because it's good, but it's still not perfect. What we've done here with the AI, because security analysts still don't fully trust the recommendations, is we've given all of the evidence that the AI pulled back that the analysts can go look at.
If I go drill into these findings, now what I get is all of the information that these multiple agents pulled back that a person wouldn't have to do. Or if they did do it, this would take hours or days to do in an organization. You would typically have four or five people. One person would go look at the threat intel. One person would go look at firewall logs. This activity would take multiple days all the time data is leaving the environment. Using Agentic AI, we've done that in seconds, but returned all of the evidence back. A customer can come down here, and they can verify, well, I saw some suspicious file activity. In this case, there was a script that was disguised as a PDF out on the internet.
You don't have to be a security person to know that if a script is disguising itself as a PDF file, that's probably suspicious. It's probably not legitimate. You can keep coming down here, and you can see all of this evidence that comes back. Now the really cool part, which is, and I'm going to test you for it, can we understand in 30 seconds or less what happened? What you do is if you come back to this graph, what we've done is we've taken all of the indicators, and we've overlaid them on a timeline here from the earliest activity and mapped it back to MITRE Tactics and Techniques. MITRE is a framework that security professionals use to describe what happened in an environment.
We've mapped these back to a timeline and a set of MITRE Tactics and Techniques and used nouns and verbs, which is how you and I communicate with each other to connect the dots to describe what's happening. Now I'm going to walk you through it, and we're going to see if you can understand when I walk you through it just by clicking on this. I'm going to click on the execution stage, the first stage.
And what happens n ow, I just highlight that portion of the attack chain that is related to execution and read it. Here we have user Alex, who accessed this device, Mozzarella, and user Alex downloaded a suspicious file, Factura, hosted on a malicious domain, dninfo.com, that was hosted on this suspicious IP address. Nouns and verbs. What happened? Let me hit the next button. Now user Alex accessed that same Mozzarella device, and he ran a PowerShell script. PowerShell script is not malicious, but on those findings, what we saw was that that PowerShell ran some base64 encoded information that it downloaded from the internet. People do not pass base64 encoded information to PowerShell scripts. User Alex runs PowerShell, and it connected to that same IP address. This happened a couple of hours after the first execution event.
At 18:58 P.M. to 18:59 P.M., one second later, PowerShell now invoked a RegServices script running. Again, just read it. User Alex accessed Mozzarella, that executed PowerShell, that executed RegServices. Shortly after, RegServices connected to a different IP address. For you non-security people, this is indicative of command and control activity. There is an IP address out on the internet. My machine is connecting to it. I have never seen my machine connect to it before. Some PowerShell processor connecting to it. This does not look right. If you just follow the chain, you have some long-lived connections, some command and control, and then ultimately, user Alex from device Mozzarella exfiltrated data out to that same IP address.
That attack chain, being able to step through it, we believe will greatly accelerate the understanding of an incident and give folks the ability to describe it to their boss, the manager of their security operations team, maybe their CEO, maybe their board of directors, or an auditor who might be coming in and asking some very hard questions. We tied it all back to Meraki and leveraged the power of the network to really drive these findings. Now, before I call Kamran up, show of hands, how many people could generally follow along with that and feel like they could explain it to somebody? You guys are kind of shy, bashful. More people than drove Teslas, so that's probably a good thing. This is in early access. It's in alpha right now.
It will be coming out in early access in Q4 with general availability in Cisco's Q1, so kind of the September, October timeframe. Like I said, the most important thing from our perspective is that we are fully leveraging the power of the Cisco infrastructure because lateral movement, by definition, involves the network. Cisco shines when security meets the network, and we are melting security into the network with products like XDR integrated with Meraki. Now, with that, what I want to do is I want to call Kamran Siddique up. He's the CISO from Steve Madden. Steve Madden's one of our most valuable and strategic partners. You just told me 117 stores now with something like 3,000 Meraki devices, I think, sending data into XDR.
Exactly. 100% of the fleet.
That's amazing. We have three questions. I'm going to ask Kamran. We're going to talk about kind of a little bit of the past, a little bit of the present, and a little bit of the future about why he chose XDR, how he's getting value out of XDR, and kind of what he's thinking about next. If you think about just the scale and complexity of Steve Madden, and I have to admit, I've never heard of Steve Madden before I met Kamran. I'm not a brand person. My wife told me that they make shoes or something like that, and now I know all about Steve Madden. When you think about the scale and complexity of your retail organization, what specific security challenges were you faced or having?
What was it about Cisco XDR that you felt like could help you address those security challenges?
I started at Steve Madden about two years ago, and we had a bunch of tools. I joined as CISO and head of infrastructure, and we had more than 40 different tools, very fragmented visibility, not a lot of, especially from stores' retail perspective, very old hardware, more than 10-12 years old, and no netflow or logging coming in. Basically, we were driving blind. What we did was we did a project where we rolled out Meraki, everything from MG to MXs to switches, the whole Meraki fleet, Meraki network gear into stores. XDR was kind of the secret sauce which brought everything together. Responding to security, the more telemetry you have, the more correlation you can build, the more confidently you can respond to security issues.
That was the driving factor for us, that I wanted something that I can send as much data as possible from all of the different telemetry sources, and then that system should be able to correlate and give me what matters most to address.
Awesome. That's the past. We kind of talk about the next one. Beyond, like we talked about native integration, obviously with all of the third parties at CrowdStrike and Palo Alto and Meraki, were there any specific features of XDR that kind of had the biggest impact on your security operations?
Yeah. I think first thing, the ease of integrations. You said it earlier, Cisco integrated with everything Cisco only. With XDR, we were able to integrate. Right now, we have almost two dozen integrations into XDR, and it did not take rocket scientists to do those integrations. It was pretty fast. One thing is speed to integration. Second thing is how well does it correlate? How does it help my analysts? We are retail. We are very cost-conscious. I cannot go and ask for 15 SOC analysts to look at alerts. For us, it was ultra important that what matters most, it gets escalated to a human for responses.
As you think about the future, how do you see Cisco XDR maybe evolving? Here is your chance. What would you like to see Cisco XDR do to kind of help evolve with Steve Madden, with you, as you are thinking about evolving the security posture and increasing the overall security posture of your organization?
I think really important, more telemetry, better correlations. If you look at AI, we are looking at how to use AI to defend. Attackers are looking at AI to have more sophisticated and more persistent attacks that they can launch. For us, it's ultra important that we can respond to all these attacks at machine speed and at machine volume. It's ultra important for future that we get kind of more correlations, more telemetry, and then kind of more actionable alerts that go to the analysts and automation. I want to be able to confidently say that this one, I'm OK for AI to drive my kids home, so respond to this alert. We have some of those already in, for example, Defender. We configured that if this behavior is observed, quarantine this device.
I want to be able to give the same confidence to XDR to be able to take actions on our behalf.
That's great. I'm going to throw one kind of bonus question in here for you, maybe not prepped for. If there's kind of one thing that you wanted the folks in the audience here to kind of know and think about, maybe they hadn't heard about Cisco XDR, the one thing that you would tell them that you didn't know and you wish you had known at the time, what would it be?
I think probably the forensics piece that is coming out. We went through a security event. For us to understand where the attack started, what path it took, how did it morph, when it started connecting to command and control, I think that's going to be very valuable to kind of stitch that story together faster. We ended up engaging Talos IR team to understand that. I think that capability will kind of enable us to be able to do the same.
Yeah. That's great. That's a great plug because while Talos is kind of feeding all of the, and it's all of the fuel that goes into XDR from a detection perspective, at the end of the day, you can't ever expect your vendor is going to be able to detect everything on your behalf, and you're going to need the tools to be able to go in and figure it out. I didn't show you guys the XDR forensics capability, but it is available on demo over there. If you think back to this kind of attack that I walked you through, the one thing that I didn't show you is how did that organization get exploited to begin with? What was the initial access vector?
Using the XDR forensics capability, you can see the email that came in to that user, Alex, that caused him to click on something that took him out to the internet and ultimately downloaded that script. That is the capability that Kamran was talking about.
Kamran, thank you so much.
Thank you.
Look forward to continuing to work with you. Everybody, thank you so much. I know it was short. It was quick. We're very excited about Cisco XDR. I know a lot of you might not be security folks, but really what I want you to take away is Cisco is melting security into the fabric of the network because Cisco shines when security meets the network.
For all of you Networking folks out there who might have said, hey, this is a security conversation, that's somebody else on my team, or somebody else needs to handle it, I would encourage you, leveraging the power of Meraki and the integration with the Cisco network, to go back and see if this is something that you feel like your teams could take on or your customers might be able to take on with this new capability around XDR that really greatly simplifies the understanding of attacks and gives recommendations on how to act. If you have any questions, you can find me. I would love to talk to you more about it. Catch Tom and Mike's presentation tomorrow afternoon, I believe, from 1:00 P.M. to 2:00 P.M.
Tom Gillis, our SVP of Security, and Mike Horn, the SVP of Security for Splunk, will be talking about how Agentic AI is being weaved throughout the Cisco security portfolio, inclusive of Splunk security, and how we're just doubling down on all of the benefits, some of which you saw in Cisco XDR today. Thank you so much.
All right, everybody, welcome back from another fantastic center stage session. This time, we just heard how Steve Madden Limited is accelerating their threat response utilizing Cisco XDR. All right, why is this so vital for you? As we just heard, in today's threat landscape, our security teams are under this incredible quantity of pressure to do more with less. You know this. You know exactly what it feels like. You deal with it every day. Fewer people, more alerts, constantly scaling, constantly compounding complexities across our IT environments. How do we keep up? The fact is traditional detection and response tools, they just don't cut it any longer. Sad truth, they are just not up to the task. They can't deliver the speed, the visibility, the simplicity that our modern teams need right now.
As we just heard from AJ, our VP of Product Management, and from Kamran, Global CISO and Head of IT infrastructure at Steve Madden, Cisco XDR is changing the paradigm. It is changing the game for security operations. We just got to see firsthand that Madden journey to a more efficient, more effective SecOps model. We saw how Cisco XDR helped Steve Madden to reduce noise, to accelerate investigations, to improve their outcomes. Another fantastic center stage session. We're so glad you got to see it. I'm Steve Multer. Our entire Cisco TV team is so glad to have you here with us on the live broadcast coming to you from Cisco Live 2025 San Diego. Quick reminder, keep reaching out to us on social media. Keep those posts going, whatever platform you prefer.
Just make sure that you incorporate #CiscoLive and @CiscoLive so our social media team can check out what's going on. Right now, we're going to head back out into the WAS where Robb Boyd is hanging out with Kevin Kennedy talking more Cisco XDR. Hey there, pal.
Steve, I love it when you say the name of my guest as well. This is one of the ones that I was going to nail, though. Kevin Kennedy, VP of Product Marketing.
You got it.
Oh, let's mark that as the first time I got it right. Kevin, welcome. Thank you so much. I am curious because XDR kind of snuck up on me as a very Cisco-focused person, but it was 2023, around about, I believe. What was the, forget the term, catalyst for why XDR, why did we get into this space, and what are we doing with it?
Yeah. I think we saw three things. One was there was a need in the market for network-native XDR. A lot of XDR comes from the endpoint side. That's good. Endpoint's a good signal. Now attackers land on the user device at the perimeter, and then they move laterally into the data center.
It's that lateral movement that really caught us off guard for a while. XDR, just for anybody, is extended detection and response.
Detection and response, yes.
OK, just making sure I had that right.
Yeah.
OK, so addressing lateral movement, new attack vectors.
Lateral movement, network is the name of the game. Who better to do lateral movement than Cisco and network better than Cisco? The second thing was many of the XDRs are closed. They're proprietary. They treat the first-party products as the king, and everything else is kind of second class. We see the adversary as outside, not other vendors. Our goal is to stop them. We treat all of the products as first class. We integrate with CrowdStrike. We integrate with Microsoft. We integrate with Proofpoint on email security, as well as the Cisco products. That was the second thing we saw. The third thing we saw is that there is a tremendous need for something that is simple for teams that are just starting their detection and response journey to get time to value.
We designed this from the outset to be easy to use, to get that speed to value.
Would you say this is a good place for customers to start if maybe they would not say, I have a SOC? They are a little afraid to say, I have a SOC. I have some people that need help. This provides a simplified way for them to maybe get control of some of that. S cale as well.
Yeah, this is often. Yeah. This is often teams that are smaller to start, and then it gives them room to grow. Even in the partnership or the integration with Splunk, which is on the other side here, we can then plug in and continue that journey into a full SIM and a full unified threat detection and response system.
I like that. There was a conscious effort to say, we're not building something that is very Cisco-oriented. You're something that's saying, we're going to treat these all equally. It probably gives you more options in more situations, not just to scale up small, but start with what you have and then work from there.
Yes. We love to meet customers where they are. That's a consistent theme across security.
I was fascinated just doing some of my pre-research around some of the innovations you've had. One of the ones that sticks in my mind, because I write a lot, is storyboarding and just the fact that you can tell us the story behind an attack, if I'm even saying that right. I know there are others. Walk me through those.
Let's start with one, which is actually kind of on your left shoulder. Meraki. One of the, there's tremendous value in this network telemetry. One of the challenges is getting access to that telemetry, especially in a large distributed enterprise. We, working with the Meraki team, did a native integration. You can turn any Meraki MX into a sensor in 30 seconds or less.
Oh, really?
Now organizations with hundreds of these MXs across their environment can get immediate visibility. It's a game changer for those types of organizations. This has really helped to fuel a major growth in XDR.
That's interesting. OK, so Meraki is a new integration. What other stuff are you promoting?
The other, which is much easier to show than to explain, is attack storyboard. I talked about making this easy for teams. Our goal with this, and this is brand new. We just launched it at RSA, is to make it possible for anyone, even a non-security professional, to understand an incident in 30 seconds or less. That's really hard to do. If you look at this, it's pretty simple, which is I have a user, Alex, who accessed a device. They downloaded it. They accessed a suspicious file that went to, that triggered access to a domain. They downloaded that suspicious file. They launched a PowerShell process that had a suspicious process, again, accessing that IP address, and then exfiltrated some data.
You can very, very quickly get a view of exactly what happened with a timeline mapped to MITRE tactics and techniques.
I like the tie-in to MITRE because you have this notion of how do I decide what's most important when I'm flooded with so many alerts? You guys are helping visualize that. Does that lead us, what else have we got to talk about?
Yeah. So one of the other things, and this goes into the theme of the show, Agentic AI and what's happening in the world writ large. We see a tremendous opportunity to change the way that security operations teams work, leveraging that. We are the first vendor. We have a feature called incident attack verification. Again, we launched this at RSA. This is using Agentic AI. You can think of it as, you said, these teams are small. They're getting started. They don't yet have a SOC. We can give them a SOC with agents. Think, one, this is SOC manager, one that is specialized in doing forensic work, endpoint forensics, one that knows how to pull Splunk data, one that, so trained for those specialties.
I have employees. I have my agents.
Exactly. Now that SOC manager says, I have an incident. I look at it. I have the summary. Now let me determine a work plan. Do all of the execution. Come back and say, is this malicious or not? Yeah, with very high confidence. Now you can trigger a reaction at the speed of AI rather than waiting for people to dig in and spend hours potentially investigating and coming to a conclusion.
It feels like when you lay stuff out like this, I also feel like there's an educational component to being able to explain. Because that part of that kind of thing, especially initially when you start in things like this, I think the general reaction in my experience has been, oh, wow, we have a lot of things to chase. As you baseline and you get integrated, it begins to calm down a little bit. You can really focus on what are the most outstanding things you're constantly tweaking and kind of getting that better. Beyond that, is there anything else from a new perspective we should bring up?
Those are the two key ones we wanted to hit on.
There is a final thing here before we go. Where do you see things going with AI driving? What is what you are looking forward to announce next, perhaps, that you can share with us? Maybe in a general sense. You do not have to give away the whole store.
I think we continue to look at how to broaden the ecosystem of products that integrate so that we can meet more and more customers where they are. We continue to look at ways to use AI to leverage and speed, automate the SOC process. We look for ways to integrate deeper and deeper with Splunk so that customers can get value out of the combined solution to really develop and implement their SOC of the future.
So much fun playing with the data and all these different ways. Kevin, thank you so much. I appreciate your time. Guys, just before we throw it back to the studio, I just want to point out, this little gradient that they've got across the top here, you'll notice this across the booths featuring security, as I've been told. I think that's one of the keys to how you can.
This is light.
In this over here. Guys, there's other sides. We've got Splunk on the other side. This is all underneath. Julie, if you turn around and look up, I don't know what kind of bad angle that's going to be. We are under the digital resilience circle, the big circle of intelligence that we have here. Steve, I think I'm throwing it back to you at the studio. If you're there, you're ready.
No, you were exactly right, Robb. You're bringing it back to me over here. By the way, I'm so glad that you described the color coding. I didn't really know what was going on with that.
It just ties together down in the showcase. That's perfect. Thanks so much for that. Thanks so much to Kevin as well. You know, as we talk about this XDR, this extended detection and response story, so much of it is based on partnership. We just got to hear that from Steve Madden a moment ago. We are going to go into another partner story right now. Partnership is everything to us at Cisco. We do not get anything done without our partners, without people talking to us and telling us what they want to accomplish, what they want to do, bringing challenges and working together side by side with us. Oliver said it in the keynote this morning. He said that the fundamental principle for us here at Cisco, everything starts with you. All of what we do here is built based on your feedback.
It's based on you driving us. That's whether you're a customer. That's whether you're an internal employee with us watching the broadcast here at Cisco. Certainly when it comes to our customers. One of those customers, Patrick Milligan, CISO at Ford Motor Company this morning in the keynote, he said, ask yourself, who do you want to work with? Who do you trust? People tend to trust Cisco. Why? It's what you're hearing about all week here at the show at Cisco Live. We are going to extend that trust conversation, that partnership conversation right now. I'm going to send it over to Richard Piasentin, our VP of Provider Connectivity Assurance here at Cisco. Richard, I'm going to let you take this next partner story away.
Great. Thank you very much, Steve. It's a pleasure to be here. I have the pleasure of being responsible, as Steve mentioned, for Provider Connectivity Assurance. Many of you would potentially know us from our prior name, formerly Exceedion. We're responsible for the observability and AI ops functions for Cisco critical networking. You would have heard Jeetu in several of the keynotes and other executives talking about the importance of observability generally in Cisco's architectures and in our strategy. Very specifically, PCA is used in critical networks. You might ask yourself, what's a critical network? A critical network is the type of network that isn't checked, but rather it's a network that must be managed in real time.
A critical network is a network in which minutes are an eternity, seconds are a luxury, microseconds and milliseconds are where you execute the fundamental controls to make critical applications themselves function. If you're doing round-Robbin testing, as an example, every five minutes, lives are affected and endangered in critical networks. I am very pleased to introduce our guest today. We have Doctor, and he probably will not like the fact that I am introducing him that way. But he is Doctor Thomas Compa, IT Solution Architect for a company that needs no introduction, Audi. Now, Thomas, first of all, thank you very much for joining us today. I appreciate you taking the time out. I know everybody on the floor, especially in the IoT space and the OT space, they just want to talk to you and find out what you guys are doing.
The reason they want to do that is Audi is leading the industry in transforming complex manufacturing systems and really taking them to a whole new level. Tell us a little bit about just generally what you're doing with Cisco and your journey on this industry-leading, something that I learned about the non-virtual version of these in school many, many years ago, but virtual PLCs and how they're transforming the manufacturing processes for companies like Audi.
Richard, thanks for the really great introduction. What are we doing? We are transforming this industry by moving all of these, we call them IPCs, industrial PCs. These are hardened devices in the factory that have been there for, I don't know, 30, 40, 50 years. They're all hanging there around, thousands of these. We have, as you just mentioned, the programmable logic controllers, or in short, PLCs, that are controlling the robots that we are having there to manufacture the cars. Anyone who wanted to talk about cars here, sorry, I have to, yeah, that won't be the topic today. It's about the manufacturing of the cars, which is interesting too, right? Here what we are doing there, we're taking all of this hardware that we have in our factories, this joint everywhere, all at once.
We're taking it all here into an edge cloud platform. This is an on-premise cloud, private environment. There we are virtualizing previously hardware-based applications onto a consolidated platform. Now if you think about moving stuff from A to B and B has still some kind of connection or relationship to A, we need some glue to hold it all together. This is where networking comes into play, where Cisco comes into play, and how we started collaborating more in our long-standing relationship.
That's great.
Maybe one more thing. You mentioned VPLCs. It is basically like the, I would say, this cutting-edge technology that we are bringing here into the shop floor or from the shop floor away into our data centers. That requires not only reliable networking, but also microseconds, just as you mentioned, microsecond latency. It is just the first application of many that will require that. We heard lots about AI, et cetera. All of these applications require just the same in terms of latency, et cetera. Also the assurance that when if something fails and you know things fail all the time, we want to know what broke and how we can fix it.
Yeah, makes a ton of sense. Maybe to put it in context for the audience and to understand why this falls under that definition of critical networks that I opened with. These AGVs, these manufacturing tooling, if you will, I mean, these are massive pieces of machinery. And you have humans that are intersprinkled amongst these machines as they're executing the manufacturing for Audi. If you aren't managing that network in real time with cutting-edge observability, as an example, you are putting people's lives at risk, which is, of course, is unacceptable. Let's look at the positive, though. What happens for Audi from either an economic? What's the driver for doing it?
By decoupling our previously hardware-based applications onto a consolidated, we call it hyper-converged infrastructure, we can easily scale up and down how demand drives us. We can have more or less applications. We can increase the size or size it down. There are obviously the sustainability goals that we have to drive, carbon footprint, et cetera, energy that we want to reduce. All of this is contributing to this vision that we are driving here. At the same time, we have all the data now already in a data center. Before, it was all there in machines and stuff, hard to get to because.
Sneaker network.
Right, exactly. You want to separate the network from everything else. This is highly secured. Now you have everything already in a cloud environment. You can do all the fancy stuff that everyone's talking about, optimizing processes, running some kind of computer vision, et cetera. It is much easier now to onboard new use cases with the platform in place.
I mean, you're really driving tremendous velocity increase in your ability to flexibly manage the space that is essentially the heart and soul of shareholder value for Audi. You're doing it with Cisco. I am happy to say, as the PCA guy, that we're partnering with you to give you that real-time observability that you need in order to manage these very, very important and very complex critical networks. Thomas, if you had one more thought to perhaps leave us with, whether that's Cisco, Cisco's partner community, other customers out there, what would be your recommendation on how to start thinking about critical networks?
First of all, for people who are listening, especially in our industry, it does not have to be automotive, but in general, manufacturing, pharma.
Theme parks.
Theme parks, yeah. Disney, great job on that one. I heavily recommend.
No plugs.
Yeah, no plugs whatsoever. I really heavily recommend at what we're doing and how you could adopt it in your future because we firmly believe this is the right thing to do for our shared goals as a c ommunity, right, but also for your own personal goals. As for critical infrastructure, yeah, we have critical infrastructure at Audi in this case because one small interruption forces us to stop our processes.
Not too good.
Right? Yeah. Exactly, exactly. Not only you, but also your colleagues from Cisco, everyone, right?
Thomas, if I may.
Yes, go ahead.
I think we all do have that responsibility. I think that's a great spot, actually, to say thank you.
Thank you. Yes.
Appreciate it. Steve, I'm going to throw it back to you.
Richard, I so appreciate it. Thomas, such a great story. Thank you for the partnership again, as Richard just said. Richard, great job. This is exactly what we need to bring, not only to our own people, but to the industry so we can get that inspiration. Thank you both. Truly appreciate it. We're going to head back to center stage. We're going to talk about experiences amplified. Specifically, how can AI fuel better employee, better customer, better IT experiences for our future-proofed workplaces? The fact is, AI is transforming the workplace. It's at a ridiculous rate. It's also enabling seamless collaboration, personalized customer interactions, automating routine tasks. Let's send it out to Snorre Kjesbu and Jay Patel out at center stage. Here we go.
W elcome. Good to see you. And good afternoon. It's really great to be here in San Diego with all of you. Last year, we actually did a dance together on stage. We're not doing it this year. But really good to be back here. What I'm going to do and what Jay and I are going to do today is really to build on yesterday's keynote. It's really about making AI real. If you look at what you have up behind me here, this is Cisco's blueprint. It's about AR-ready data center. It is about the future-proofed workplaces. All of that is built on a secure global connectivity and digital resilience. If you see that, everything, the entire story is coming together as one story. All of this is accelerated by Cisco AI. It really delivers a truly integrated experience.
It is not about tech only. It is also about how you empower, how you lead, and how you do this in an AI-driven world. What does this mean for collaboration? What does it mean for future-proofed workplaces? We have the traditional building blocks that you have seen. It is about networking, how that powers connecting everyone, security to ensure the safety and resilience, observability. You will hear a lot more about the manageability, the observability as we go along, when it comes to how it provides visibility, how it provides insight that actually can be used in all of these setups that are now mission-critical, and collaboration that elevates how people work together and how we actually get things done. In an AI-first, human interaction matters more than ever. It is all about the experience.
It is really all about being able to bring people together to get that engagement going, to get your customers to engage in different ways. Not only that, but when you are working with the experience, you need that right experience there so that people are drawn in and they can use it. Because a good experience and a bad experience is a difference between getting things done in a meeting or when you are looking at a Contact Center actually losing a customer. It is absolutely essential. Your customers expect to be able to interface and work in different ways and really get there. We need to raise the bar for human interaction. For years, we have been delivering on exceptional customer and employee experiences across our portfolio.
If you look at what we have up there with all the things we've done, we've been working with NVIDIA for a decade. Before anyone knew who NVIDIA was, it was about really doing acceleration of computer gaming. Now they're really the core company on the product. All of these things have taken us to the next level. It is a foundation for how we truly create intelligent experiences. At Cisco, it isn't new. We have been innovating. It is a long time since it became the industry's hottest topic to make it work together, to interact more seamlessly, intuitive. What it's really about is that you get AI to help you in the way that you think, you adapt, and you act. What we are looking at is we're focusing on that proactive engagement.
We're focusing on that much rather than that being just the assistance as well. Look at the things we've done. We are amplifying a lot of our features using AI. Look up here. You have the AI coder that handles extreme packet loss. We have the AI assistant that helps with meeting summaries. It helps with highlights. It helps with all the things you're doing in the meetings. You have conversation summaries for Contact Center agents. If you look at these common denominators, you see they go across. The things that are useful for calling that are useful for meetings are also useful for Contact Center. We're building that across. We have a ton of new devices that are optimized for AI and so much more.
I know that you out here, and I know a lot of you, and I see a lot of familiar faces. You are all about experience matter. You share our belief. In the past year, we've been working even more on creating even more exceptional experiences. What does those include? It includes things like improved audio and video experiences, advanced AI tools that can actually help you with immersive and human interaction. You can do that across this entire suite, the collaboration devices, and also Customer Experience. I think one of the inspirations for all of this, and I almost think of her as a friend, and that is Bobby. Bobby, and you're going to meet Bobby now, she is the unsung hero of IT. She is facing some very interesting challenges on a daily basis.
I would like you to meet Bobby and see what a day is like for her.
This is Bobby. Every day, she deals with the impossible.
CEO is looking for you.
What's broken?
The call with investors, of course.
What part isn't working?
I don't know.
The investor meeting.
It's the network.
Probably the router.
Always the software.
It's user error.
Are you sure?
Trust me. It's probably maybe user error.
I'm already on the VPN, but I want to be a VIP.
You need a new system. Something with AI, right?
AI is good.
We need it like last week.
This doesn't count as booking a conference room.
AI chairs, desk, plant. AI plant.
You do know what AI is, right?
You're going to love this.
I have solved an age-old problem and created true interoperability.
We can just jump to another one whenever one fails or another stutters or a third does whatever.
Can I install Minecraft on my laptop?
How come I can't use Tinder for business communication?
I had to log into five different programs to answer one question. I quit.
Give me something future-proof with self-service and a conversational AI that actually works for our customers and our agents.
I know it's a company computer, but it was kind of just lying there. Can I mine crypto with this?
My dog died, and his name was my password. I am going to need to change my password.
Any resemblance? Do you get any of these requests every day? You know, for me, it really strikes a tone and a note. What we want to do now is that how can Cisco help Bobby solve for these problems and all of the things that she's in the middle of? To kick us off, I would like to invite my friend and great colleague, Jay Patel, to talk about Customer Experience. Jay.
Hi, Snorre. Thanks very much, Snorre. I really love that video. I am Jay Patel. I run the Webex Customer Experience Solutions Group. We all know that experience matters. We all personally also know just how important Customer Experience is. We know that a third of customers will leave a brand after one bad experience. We know that brands that focus on revenue and CX increase revenue and reduce costs. Yet only 25%—a quarter of customers—are actually happy with their last customer interaction. That is because getting CX right is actually very difficult. Firstly, all of us have raised our expectations. Meeting them is getting harder. There are more channels of communication. There are organizational complexities. There are silos. There are security issues. We believe, though, that AI is a game changer. It gives us the tools for delivering consistent, great CX.
Now, I know that many of you probably are in this cohort of two-thirds of companies that are looking at AI for customer support. For those of you who aren't, let me explain why we believe that CX is the killer application of AI. If you look at what's been happening in the last four or five years, firstly, the models, and I think hopefully most of you saw Kevin Weil from OpenAI yesterday, the models are 10,000x what they were five years ago. We also now talk about investment in this sector, not in millions or billions, but actually trillions of dollars. What this means is that AI is rapidly solving the problems that many of you have had in Contact Centers. AI can now understand intent.
The background noise, accents, missing words, ASR has gotten so good that accuracy now will understand intent, even in the noisiest of environments. I can now also generate natural language to allow you to have a less robotic conversation. I can also help you hold real conversations. You have models for intonation, for turn of speech, for interruption. I can also help your developers with the code they need to integrate all the back-end systems. I can also help you provide real-time data for continuous monitoring and coaching. What all that means overall is that you can now deliver true personalization and accuracy at scale. It allows you to create virtual agents that are indistinguishable from human agents. It ultimately allows you to resolve problems, not just create tickets.
It's why, again, I mean, you've probably seen this on many slides, but for me, at least in my area, CX and AI coming together is one of the largest, most profound technology opportunities of our lifetimes. It's why we've embedded AI across our entire portfolio. Within our Webex Connect product, which deals with Proactive Communication, we have AI to help you generate code. That code helps you integrate systems faster. That helps you communicate quicker. We have AI embedded in our Webex AI agent where you can build natural language and voice and digital agents. We have AI in our Webex Contact Center product, empowering human agents to be more productive. In the platform, as Snorre mentioned, we benefit from having AI being built across the Cisco platform, and we benefit from things like noise removal and translation that are happening in other groups.
Let's just go back to this Webex AI Agent and see that in action.
Imagine a world where your business can answer your customer's needs on their terms. Webex AI Agent is a powerful solution designed to reimagine your customer interactions. AI agents can automate tasks across your business, from handling internal HR and IT requests to generating sales and responding to customer service needs. AI agents can autonomously answer questions and complete actions in a way that feels natural and conversational. If the conversation heads in a direction outside the agent's scope, reliable standards are in place to keep the conversation within the guardrails set by your organization. Webex AI Agent also scales effortlessly, supporting multiple languages across voice and digital channels. Building and customizing your AI agent is all possible through an intuitive design tool with the ability to build simple to complex use cases.
You can also integrate your AI agent with your critical backend, like your CRM, ERP, and HR systems, to power real-time solutions for your customers. Scalable and consistent experiences that your customers trust. That's the new era of customer engagement with Webex AI agent.
Look, what you saw there was a capability for a business to have an always-on intelligent front door 24/7. What that allows you to do is ultimately meet the expectations that have risen. It allows you to scale your CX operations. It allows you to have natural human-like interactions, fulfill intents by integrations to back-end systems. Ultimately, it frees up your human agents to deal with more complex issues and spend more time on them. Today, it supports nine languages. It will support more going forwards. You have security and privacy built in by design. It is GA for voice and digital channels today. What we are also doing is we are enhancing this AI agent continuously. I am pleased to announce that we have just launched some pre-built templates. This will allow you to sort of quickly create agents for common and industry-specific use cases.
Because what we really want to do is we want to kickstart this revolution by giving you the tools so you can start the POCs quicker, create your business cases quicker. We are also happy to announce that we will give you the ability to choose which LLMs power your AI agent. Just look at how easy these templates are. You access the AI agent from within Control Hub. You select pre-built templates for typical use cases. The templates are pre-populated instructions and actions which can be customized. Those actions can access backend systems like ServiceNow or Salesforce or any other system you have. You can then preview the voice or digital experience to then optimize it before you launch and ultimately monitor the AI agent activity once you have created it. This really is only a first step to making it easier for you to use these agents.
The actual creation and deployment and training of agents is going to be a very complex task, not dissimilar to how we look after our employees. We will continuously improve this capability for you to be able to use this. The early customer results have been very encouraging. United Rentals, a leasing business, looking to reduce IT support calls by about 85%. It's well on its way to achieving that. Bank First has reduced its average handle times by 22%. CarShield is probably the most interesting one where ultimately they have been using the AI agent to deal with inbound calls and triaging at a first line what the call is about before they pass it on to a human agent. In the initial weeks and months, they've been able to halve the number of agents they need to actually do those frontline triaging.
What this illustrates is that you have the potential to automate repetitive interactions 24/7 and resolve some of those basic issues. Now, even though AI agents can help with Automation, we also still believe that human agents are needed and will be needed more in the future. We are also working on AI augmenting human agents. Our Cisco AI assistant will make agents more productive and empathetic, supervisors improve performance, and manage one of the biggest issues that we have in Contact Centers, which is attrition. Attrition at times can be something like 30%-40% per annum of a Contact Center and its agents. We are giving supervisors tools to manage that and also help managers improve business performance. If I look at what we have done with our Webex Contact Center and AI, we have thought about the customer journey end-to-end in a Call C enter.
We've applied AI to every part of that customer journey. One of my favorite features is actually the drop call summaries. Because often, you will make a call. For some reason, it gets dropped, and then you call back, and you have to repeat yourself. Again, with drop call summaries, we're able to transcribe that initial call, and then when you call back, send a summary to that agent so they can pick up where you left off. Again, no need to repeat yourself. During the call, we are able, because we can transcribe, we are able to think and suggest responses to the agent. When the agent is talking to you, they will get some nudges as to what they should say next. That should improve their ability to serve customers better.
Post-call, some agents may typically spend up to 10 minutes summarizing the call they've just had. By providing meeting summaries, we can reduce that because they then get the text straight off the bat, and then they just need to edit a bit. That can go from 10 minutes down to 3 minutes. For supervisors, we've got tools to manage agent stress through wellness breaks and to do performance management through auto CSAT scoring. For managers, we have things like topic analytics so they can address some of those recurring issues that you get in a call center. Cisco AI assistant, we believe, ultimately will deliver lower costs, higher CSAT, lower attrition, and ultimately more productivity. One of our banks on the West Coast here basically has been using some of these AI assistant features.
In addition to the quote that you've got up there, one quote I liked was they said, "By using these things, we've been able to meet our SLAs at a much higher percentage, even on the busier days." For me, actually, that's what we're trying to do. We're trying to augment Contact Centers when they get busy so they can maintain high levels of service. Now, again, one thing you've seen yesterday, and I think you've probably seen it in many of the talks, is that Cisco is committed to meeting businesses where they are, whether they've migrated to cloud or stayed on-premise. I'm happy to confirm that the advanced AI features that you've seen on Webex Contact Center will soon be available for on-premise customers as well.
Our new release of CCE Release 15 is a refreshed platform that's got additional digital channels and an ability to use our cloud AI tools. This will enable on-prem customers to use the Webex AI agent and the AI assistant products. Release 15 also will allow integration with third-party virtual agents. Now, in addition to all the work that my group does, in addition to all the work that I get from Snorre's group, what we also benefit from here is we're part of the Cisco platform. You heard Jeetu yesterday talk about the huge investments we're making in AI security, observability. I think when you start considering what you're going to use going forwards, one of the things I believe is that you're going to need to be secure in all of the LLMs that you use.
One of the first places you will use an LLM is probably in customer service. We benefit from being part of that protection that Jeetu and the platform teams are introducing. We believe that essentially at Cisco and at CX, we provide a system for managing AI, not just a point solution with AI in it. Ultimately, this platform advantage, we believe, gives us an unparalleled foundation for innovation. With that, I'd like to bring Snorre back up to talk about employee experience. Come on, Snorre.
Thank you, Jay. It is so good to see all of the strides we're making on the Customer Experience side. As Jay said, we are taking these components and we're using them across, and we're learning really from each other here. What you've seen is that we are really supercharging that Customer Experience as well. What we are going to do now is that we're going to dive in, and we're going to look at how is that happening for the employees, the ones that are actually out there, they're in the meeting rooms that are creating and having the discussions and needed tools that are there. What we see is that AI is the cornerstone of actually creating those exceptional experiences that we want to have there. Really, AI has become pivotal in how we do it.
It is really the difference between delight and frustration. When we talk about our approach to this, I think it is important to say that there are two key customers that we look at. When we think about employee experience, we think about these two. They are the ones that actually use the service, the ones that are sitting in the meeting rooms, the ones that are using Webex for chatting, the ones that are actually making the calls, the ones using the video devices. Then you have the other one, which is the ones that own and operate. That is the IT organization. It is the facilities organization. We actually address both.
What I'm going to do is that I'm actually going to talk about both of those to see how do we bring the experiences into both, not only the ones that use, but also the ones that own and operate. Our unique approach to this has been that we actually have software and hardware. I'm frequently asked, what is unique with Cisco? It is because we have Contact Center, meeting, calling suite, backend, app, and device front end on top of AI and the rest of the Cisco architecture. We can go beyond because we can actually then add a lot of functionality that goes across all of those. One of the things I mentioned in the beginning, a lot of list of things that we've done. One of the things I said was the Cisco AI Assistant and what it can do.
We have gone out there and we've spoken about it, and it's in use all along. The adoption is really widespread. We are evolving those capabilities really, really rapidly. We see that it's a game changer for productivity at scale. Personally, I'm never in a meeting anymore where we don't transcribe, where we don't automatically take actions, where if you're late into that meeting, you're actually then being caught up. We had a briefing on Friday where one of the analysts arrived 10 minutes late into that meeting. She was brought up, and she said, "Wow, this really helped her as well." What it helps with, it makes it smarter, it makes it faster, and it makes everything more intuitive. By adapting to that, we are rapidly changing how employees work, and we are more flexible, and you're more agile.
We are announcing one more thing, and that is what we call the AI assistant scheduler. Normally, when you're scheduling a meeting, you need to understand the context, you need to understand the scheduler, you need to understand who you're going to invite, you need to find time, you need to make a title and agenda. Now, with the Cisco AI assistant, we can actually do that. If you click on the CTA button, you can actually get the action items. You can get the proposal of what that meeting is going to contain. You get a draft schedule for that meeting. There is still a human in the mix. When you've gotten all of that, you can literally press the button, and the AI assistant will actually schedule that meeting for you.
It has taken out all of those other steps that you had to do because it is basing that on the context, and then you can decide whether you are going to do it or not. That is a brand new functionality that I am very, very proud of. The other thing is that Cisco is about open. It is about partnerships. Jeetu yesterday spoke about the Microsoft partnership we have, but we have a lot of partnerships. At Enterprise Connect, we spoke about how we work with Salesforce, but now we have more. One of the things that we want to make sure is that we have the workflows that go across tools and platforms that your team rely on so you can actually bring those into Webex so you can have that workflow integration.
You can then stay in the flow of work using the app that you actually use the most. The thing that we have done now is that we have Jira now in there. You can create a Jira ticket, and that will actually update with existing tickets and details. You then have the ticket there, and you can do all of this without leaving Webex. We also now integrate with Salesforce, with people that are using the Agentforce for seamless customer service as well. What you see is that these workflows are easily configured. You bring them into Webex, and you eliminate friction, and you get things done. I'm a get-things-done person, so this is what I do: faster and smarter. We go beyond workflows to create tools that support your front line of your business.
Because there is a front line there that can be in your local branch, it can be in your regional branch. That is where the brand you have actually connects with the community. These teams need tools that are seamless, that are human-centered, and have Customer Experience. Our solutions empower teams and elevate customer interaction to do that. Back at Enterprise Connect, we spoke about Webex Calling customer assists. That helps front desk staff to gain insight from calls and serve customers more effectively. Today, we are adding more to that. We are actually adding to this that you can get text cues to the customer assist, offering customers a convenient way to basically then have their way of contacting the business.
We are integrating this seamlessly with an AI receptionist so you can actually have your AI receptionist within Webex Calling so that people can actually get more work done and do not have to do those mundane tasks, but actually do things that are higher value add. Would you like to see what that looks like? Yeah, see a lot of nodding heads. There we go.
Meet AI receptionist for Webex Calling. Your always-on virtual receptionist that can answer questions, book appointments, and transfer calls so your business can boost customer satisfaction without adding staff.
Thanks for calling Florissa Dental. I'm your AI receptionist. How can I help you today?
Hi, yes, I have a toothache, and I'd like to schedule an appointment. First, I want to make sure that you accept US insurance.
Yes, we do. U.S. insurance is covered 100%. Would you like to speak to an insurance specialist, or are you ready to schedule an appointment?
I'd like to go ahead and schedule an appointment. Thank you.
Great. I'll transfer you to a nurse who will give you an initial consultation and help to schedule your appointment.
Thank you.
Deliver outstanding customer service. AI receptionist, available later this year as an add-on to Webex Calling.
Wow, isn't that something? That is actually really taking real workloads off a dental office, like you just saw there. I spoke about our unique approach. Very famous guy said, "If you're going to make truly great software, you need to make your own hardware." We do that. We take our devices that are purpose-built for every interaction. Our devices that we have for collaboration, they are designed with AI. They are designed with interoperability. They're designed with sustainability and manageability at the core and all of those. With that, we are actually combining leading audio and video intelligence to deliver that unmatched cinematic experience. When you look at this, this is not just a meeting room. It is a vision of where you connect with everyone, where everyone feels included.
In the hybrid world, meeting spaces must work seamlessly in order to bring the teams together and use them. A lot of offices are not ready. 50% of employees out there say that their office space and their meeting rooms actually fall short. That is something that we are working on. The solution is to use and have meeting room technology to work together just like your team would. The thing that we've been doing recently is to add more intelligence to all of these components. We launched the Ceiling Microphone Pro back in the fall. It has adaptive beamforming technology that will be able to focus in on the person. It has very, very fast setup because we use AI to set it up. Things that used to take days can now actually be done in minutes.
What you see is you get premium audio, but it's also much easier to use and set up, and everything is over IP. Let's take a look at how that's coming together. Wow. That product is being deployed and deployed and deployed and deployed. We needed to actually build up a second production line to face that. We introduced that, and it's been taking off. We see that that one click is key. What you saw yesterday when Jeetu was on stage is he showed the Room Vision Pro, the Room Vision PTZ camera. That's another AI-powered solution. I'm going to talk a little bit more about and show you some more about that today because that is a plug-and-play installation over Ethernet, which means that you combine power with video and controls over one single cable for easy deployment.
It can track people in real time. It can also automatically adjust for the best shot. You have extended reach. You have highlights of the active speakers and all of those things. Let's look at the film again that Jeetu used yesterday. You know what? With everything over IP, we call it one click to distance zero. Wow, is an Ethernet cable sexy, isn't it? Because we can do so much with it. Anyway, do you want a real demo of it? It's up here, actually. You want me to actually install it while I'm here? Here I have a Cisco Network component. I have a camera. Let's set it up live, and I'll show you how hard it is. They've even trusted me to do this. What I do, I actually plug it in.
If I can actually do that, because I don't have my glasses on, this is embarrassing. This is funny. It was supposed to be very easy. Here we go. You heard the click? One click. Now, as you can see, it's now actually installing. You will see that it's going through its routine. It's looking up the room. You see that it's actually going and testing it all to the end. You see the light here. It's going through that entire installation cycle. You will see that it's now up running. Now the installation is done. You see that the camera is not yet connected. We're now waiting for the final cycle for that camera to be connected in this room. Now it's connected. We install that in what? 40 seconds. Here's the beauty.
Not only did we have power over Ethernet, not only do we have video over Ethernet, like you can see here, but what I'll show you when we get up now is that we will actually also see that we have the Control Hub here. Now you see, when you see it up there, you see that the camera is now already in Control Hub. The three things that we had here were power, media, and manageability over Ethernet, everything done at once. That was the PTZ. We are doing a lot more on this front. What we have done is that we have created the Workspace Designer to help you design these rooms. To take you through there, I have got Marta on stage with me that is going to demonstrate Workspace Designer for you.
Thank you, Snorre. With Cisco Workspace Designer, you can reimagine any workspace, from the smallest of huddle rooms to the largest, most complex configurations. You can select from pre-built dimensions or customize your own, as well as customize the furniture in the room and select your software experience. It is when you start selecting video conferencing equipment that the tool really starts to shine. We can provide you with tips and recommendations along the way based on your selections. You see now a little red bubble has appeared under tips. It lets me know that my display coverage is not optimal. This is such an easy fix. Let me just go ahead and readjust the screen size to a bigger screen. Now the bubble is back to green, and everything can be seen from inside of the room. Okay.
We all know that great video starts with great audio. The table microphones that I've selected for this space, they're not providing enough coverage for the remote participants to hear everyone in the room. This is such an easy fix. Let me add the Ceiling Microphone Pro. Now the coverage is great, and everyone can be heard from inside of the room. We're nearly there. I'm just going to select an active USB-C cable for my screen sharing option. Of course, I want to elevate the experience by adding extended speaker view with the new Cisco Room Vision PTZ camera. I also want to show you the new cable map here so you can see how radically simple these rooms are to deploy.
With that, I have a blueprint of my space, which is a summary of the workspace that I reimagined, complete with technical specifications such as equipment and coverage areas that I can go ahead and share with my key stakeholders so that we can make this reimagined workplace a reality together. Thank you very much.
Thank you, Marta. That tool is so useful. It is so useful when you plan it because then you know exactly what type of gear you need. You do not add too many microphones or too few, but you get it just right to be there. These capabilities are actually amplifying the experience for the ones that own and operate all of the services. That is IT, HR, and facilities. Let me dive into what we are doing to actually help the ones that own and operate. What we will see here is that Cisco collaboration devices and Webex Suite are built on the foundation of the rest of Cisco. It is the networking, the security, the observability. We take all of that in, and we bring in the real-time management tools. We bring in the ThousandEyes.
You get the full stack's observability, which means that an IT team of five can actually do 300 rooms without a problem. This is exactly what we're doing and that we've done out there. If we dive into this further and you look at all of the devices, and when I say all of the devices, that's also our wireless access points. It's our network gear. It is our security cameras. It's the video devices. It's the phones. Everything are sensors on our network. With that, you get valuable information about temperature, about occupation, about CO2 level, about a lot of other things that you actually bring into that. Now we are actually bringing some of this uniqueness together by combining the C9350 smart switch and the Cisco Desk Phone 9800 series. What we're doing that is actually make them work more efficiently together.
That means that you can have a centralized energy management, and you use the Catalyst Center to do that with the new 9350 switch and the Cisco 9800 series phones together. This provides a central visibility, insight, and automation. That leads to operational efficiency and eventually to cost saving. That is a lot of talking. Let's do a demo and see how all of this works. What I have over here on the board is that I actually have Cisco Spaces running here on the board. I've connected into San Francisco, our fourth floor. You see that currently at the fourth floor, 11, 12 actually now, rooms are available. Nine are occupied. You can look at the floor occupancy is relatively low. Indoor air quality is great. Temperature is 72 degrees Fahrenheit. That is warm for San Francisco. You have the map here.
Let's pretend that I'm actually coming into this. I need a meeting room for the day. I'd really like Jay. You know what? I'll select Jay. Jay comes up here. It's available. It will take 15 people, 14 people. You see the details for Jay here with air quality, et cetera. I can now hold that room for three minutes, which gives me three minutes to walk from where I am and over to the room. I can even press get directions. I have directions from where I'm at to the room. I can even bring it up on my phone using the QR code that you see there. All of that is integrated into what we do with spaces. Now let's pretend I got to Jay. I do like this. Here are my meetings.
What I'm going to do now is basically to do this and join this meeting that we have up here. That meeting has already been running for a little while.
This meeting is being transcribed and summarized.
It's being transcribed and summarized like it always happens. I have my great friends here that are joining in. What I can do now is that I can actually go in here. First of all, I get a summary of the meeting because I was late because I was presenting to you guys. I can also go back in and I can select here. I say, were there any action items? That's good to know. Have they given me any action items like double your sales in the next three months or something while I was out? No, there was one for Lisa. There was one for Thomas. There was one for Mary, but none for me. You see immediately how and what value we get from that. We'll say thank you to the team. Thank you for joining.
Thank you for being part of this. That is what we can do. Advance the slide here. There we are. What you see is that this is a real game changer. What we do is that out of using Control Hub, we get the multiplier effect. By using full stack collaboration, by having all of the tools, we can reduce things like power consumption because you can slow roll your building and you can slow roll your estate more. There is another thing that we are working at for all of you. That is what we call Zero-Touch Provisioning. That means that you can actually manage your devices from Control Hub. They can be pre-staged and they can be provisioned so that you can actually simplify your entire deployment process.
That is not only about saving time, but it is actually empowering IT teams to deliver strategic value. You could support your organization, and this will be a lot quicker. We have manageability. We have monitoring. What you can see here, you could customize your AI workflows. I think this is super key. This goes across Contact Center, meetings, and devices where you actually get your AI workflows up there. You integrate them in systems like Jira. You actually can use that to see what type of deployment you have. It enables rapid innovation and adaptivity of the type of tools and type of features that are being used. You have that new dashboard in Control Hub that provides AI adoption monitoring. That is so important. You can trend. I get to trends. You can track what is happening.
You can actually get all of that from one single pane of glass. I have one more thing I want to show you at the end here. With Cisco, we are cloud first, but we also have an on-prem focus. The thing we're doing is we're looking at hybrid versions of that with resilience. What we are doing is that we're bringing advanced AI also to the on-premise. That is. We do that to the meetings. We offer features like noise suppression, background blur for focus, and really secure experiences. We can also do things like in sensitive environments, you can then use your private meetings to keep traffic within your network. We can have on-premise storage for recording and encryption. We can make sure that your data and you have control of your data and you can be compliant.
For mission-critical operations, where downtime is not an option, we have site survivability for cloud meetings that makes sure you can actually seamlessly go and you can continue during an outage. We deliver this with security, reliability, and resilience. I know that was a lot. It is a lot to take in. If I am going to take you away from here with four key points, we bring everything together: Contact Center, meeting, calling, messaging, and more. We leverage AI in everything we do. It transformed the experiences for the employees and for the customers to drive better outcomes. Three, with Cisco collaboration devices, your one click, even if it was a little hard, from distance zero. With this, it is powered on a platform that is simple, secure, manageable, and it is a unified platform from Cisco.
We are passionate, like you are, about the experience, the experience for the users and for the people that own and operate. Let's bring some real users up here. I have a treat for you because I have Doss Tychicus from AIDS Healthcare. He's the CIO there. I have Monty Tan. He's the VP of Infrastructure at GetixHealth. He's also here. Together with my great friend and incredible CMO for Webex, Aruna Ravichandran, please come to the stage for a session.
Thank you, Snorra, as always.
Thank you. Come on in.
All right. We are at the tail end of the session. We have two of our esteemed customers with us. Let's get started. I would like to ask both of you to introduce yourself, tell us a little bit more about your company, your role. We'll get started with you, Dos, first.
Thank you. I'm Mohand os. I'm the Chief Information Officer for the AIDS Healthcare Foundation. We are the world's largest medical provider for HIV and AIDS. We have clinics, we have a pharmacy, we have health insurance, and also we have thrift stores to support here and there. Happy to be here.
Awesome.
Hi. My name is Monty Tan, VP over infrastructure at GetixHealth. We're a revenue cycle management company, obviously healthcare by the name. We support over 1,300 healthcare facilities, serving 15,000 physicians, as well as providing back-office services. For example, we managed over $10 billion of accounts receivables for our clients, over 2 million credit card transactions, and 80 million patient interactions. It is a major footprint in the healthcare back-office industry.
I think one of the interesting things which you shared with me is that you actually grew from $300 million to $3 billion over the last couple of years. Yeah. So it's awesome. OK. Let's get started. Again, I'll start with you. Before you got started with Webex, you had a choice, right? Like you looked at multiple different providers across the board. What were the challenges you were actually experiencing? If you're willing to share who you had before, that would be awesome. Tell us about the challenges and why did you basically choose us.
Sure. Let me tell why we picked Webex, right? First thing, we want innovation and AI. It is not just like a buzzword anymore. We want to have our goal is to have patient satisfaction and retention, patient retention. If you want to achieve these two goals, you need to have a system where it can support innovation and with the AI. That is the first thing. Because previously, I need to follow the phone, right? When my office phone, it rings, I need to run and grab it. Now phone follows me. In my mobile, it calls me. I can answer from anywhere. Technology has been changed. Phone follows me. I am not following phone. That is the first innovation and AI. Second thing, integration. Seamless integrated with CRM. You saw the demo with the Salesforce or any other CRM used. We use Salesforce.
Also with our EHR, our medical record system, our pharmacy system, Domo for the Power BI. All the integration is happening just like a miracle. Why it's so important for us, right, when we pick? When the patient calls us, we want to make sure we address by the preferred name. Because we support LGBTQ rights, right? We want to address them. We also want to make sure what's their favorite team, what's their favorite sports team, or favorite color. It will give the touch of the patient. It's not like you're talking to someone else, right? They understand. It's connected. The second thing, integration. Third thing is implementation. Three I's, not 1,000 I's. That's a Cisco product. It's three I's. Third one is definitely implementation. How fast we can implement. We are constantly growing.
Previously, I used to call the previous vendor, hey, we need to open the new clinic in less than two months. They said no. Now I do not even call Aruna because it is so easy. Implementation, we completed 3,000 phone lines in less than six months from the previous thing. It is so easy. As long as I have internet, boom, there is a phone system. I got more pressure from my present now because, hey, we are opening here and there. My stress has been reduced. The three I's, first, innovation, integration, and implementation. That is why we picked Webex.
Awesome. I think when you had mentioned to me one of the others, it was not an I, but it was also about the cloud, right? Like being able to stand up the Webex Contact Center wherever you wanted in terms of being able to provide that. What about you, Monty?
Our decision was more logical for us because of a lot of merger acquisitions. Of course, with the mergers acquisitions activity, a lot of legacy technology came with those organizations. We wanted to look at it and say, hey, we've got to consolidate. Having multiple calling platforms, having disjointed systems, and then having to consolidate reporting, Webex was the logical step. It plugged into the ecosystem and, more importantly, plugged into the roadmap that we had for the next three to five years.
Awesome. Everybody's talking about AI. We heard so much about AI innovation across our entire portfolio, both on the EX side, employee experience side, as well as on the Customer Experience side. We launched our Webex AI agent earlier part of this year. Why don't you give me a little bit more color, Monty, in terms of what are some of the AI capabilities you're actually using and how has it actually met your business need? If you have any ROI to share, that would be awesome as well.
Sure, sure. Us coming on board at this time and with this new Cisco wave is perfect timing for us because as we reinvent ourselves to be competitive, to be current with the industry, and hopefully take a step forward, we said, hey, we do not want to just take it to this next iteration. We want to be leapfrogging all the other calling platforms and all the feature sets that are available out there. We do not want to be just the industry leader. We want to be the vertical leader. We want to be the first one out there. We want to be the best out there. We want to be the more robust. AI assistant, AI agent, those are just plugs and pieces of the puzzle for that bigger picture so that we do stay competitive. We do stay ahead of that industry and our competitors.
Are there any specific feature or functionality? Because a lot of people are talking about AI capabilities. Like most of the time, it's very hard to demonstrate the value with respect to AI.
Right.
Have you guys seen anything in terms of being able to reduce the call volume or the first call rate or mean time to resolution?
Very good question. Of course, everybody is, all of us are in business. Revenue is core to all of our functions, or else we would not be in business. We are looking at this and saying, hey, it is not just an issue of reducing headcount or anything like that. It is making more operational efficiencies come out of using the AI, OK? Looking at upscaling our agents and associates to take care of that tier two call and let the AI agent or the AI assistant take care of the mundane things like, oh, I want to see what my account balance is. OK? Those definitely should be relegated to an AI bot, OK? Things like being able now to QA every recording, 100% of all agent recordings, OK, instead of 1%-5%, OK?
Those types of things are what are going to gain us that operational efficiency and leapfrog us, again, to that next level rather than just taking that iterative step.
I love that. Like when you say that you're relegating a lot of your tier one calls, like your first line support calls. The beautiful part with respect to what we launched is you can't even tell the difference between the AI agent versus a real operator.
That's right.
That has been our fundamental goal. Jay talked about that.
Exactly.
What about you?
Sure. I don't know how much time we have. I can talk for days and days about AI, right? It's so interesting. For us, how it's helping AI, that's one of the reasons we picked Cisco, right? When the person is calling, after the call, we get the short summary, call summary. It's so easy what we discussed, the agent discussed. Second thing, more important, when we transfer the call from one agent to another agent, people are getting frustrated if you ask the same question again and again, right? With this new technology, why we love so much, it will transfer what they discussed with the agent one to the agent two. They don't need to ask anything. Boom. It's like a seamless patient experience. That's what we are, ultimate our goal. We are proud to launch a new one, right?
We operate in 18 states. Every city, we have a pharmacy, we have a clinic. Everyone has their own phone numbers. I don't know. People cannot remember because sometimes they go to different clinics, different pharmacy. With the help of AI, guess what? We are going to have the we will be launching very soon one toll-free number. You call them based on AI. It will pick the phone number and validate because we are all Cisco shop. It validates personal information and transfers the call automatically to that particular home clinic. Boom. They don't need to remember 10 different phone numbers. It's the mind-blowing technology we are coming up with AI. I have a lot to talk about AI. I'm one of the big fans of AI. That's the future. It also reduced a lot of the stress from the employees, right?
It records the tone, how the employee health. Also with the AI technology, I don't know what's the technology Webex is using. It picks the right signal when on the global calls or multiple places. It picks the right range of internet so that the calls are so easy, easy to understand. Talking to AI agent, I'm from India, right? Everyone knows. I came in 1988. Still, I have a strong Indian accent. Webex AI agent can understand my means. It can understand everything else. It's been so good.
Awesome. I think we are run out of time, but I will ask you one lightning round question.
Sure.
If you're OK with it, what is your most favorite Webex feature? It could be on the employee experience side. I know both of you have calling. You have our entire collaboration suite. You can pick whether it is an AI assistant feature or it is the Webex AI agent. Does not matter. Tell me what is your number one feature.
Easy one.
If you have a product request, both the GMs are right over there.
Easy.
You can ask them too.
No. It's a feature that already exists. Snorri actually demonstrated summarization. All of us, sure, we're double, triple booked. We're jumping from meeting to meeting. Summarization. Absolutely game saver.
Awesome. What about?
For me, like we always talk about patients, right? Also, I want to talk about our own employees. The new one with the Webex, we are working with our payroll system. Employees can make a call, hey, I want vacation time approved. Seamlessly, it will talk and get the vacation approved. There is no need to remember the password, no need to open the app. Just a call, done deal. We are almost testing. For us, it's a big thing to satisfy employee because employee retention is also a key thing. With this new technology, AI helping the agent, our staff helping the patients. We are one of the very happiest customers.
Awesome. Thank you, gentlemen, for joining us today. Before we let you go, I want to actually play a final Bobby video. Can we actually roll the video?
Please ignore that. Come on in. As you can see, we need a little help here. Look, facilities wants to streamline the office, make it smarter. HR wants productive employees that actually enjoy their workplace. IT needs visibility, manageability, and probably a larger office. I want to future-proof this place.
I think we can help with that. By the way, I think I found one of your lost employees.
Sorry. Got evicted from my desk tent. Needed to get a little shut-eye. Let's talk. I have got the blueprint for the next great interoperable multi-screen, multi-platform, multi-computer, multi-webcam meeting room.
What is it you do around here?
I'm in accounting.
Awesome.
Folks, another fascinating conversation for the archives. We just explored how AI is enabling our customers, our IT employees, our users to future-proof your work spaces. I tell you, fascinating information we just heard. AI is unlocking so many possibilities, enabling your employees to collaborate from anywhere, anywhere. Did you hear me? Anywhere. By taking over routine tasks, you know those things that you could do without and make more time for more valuable tasks that you could be doing, more valuable experiences. We have AI agents that are going to be taking over those tasks for you. I want you to sit back because we have another amazing conversation. There is so much to consume, folks. I'm going to turn it over to my phenomenal Co-Host, Steve, who has another amazing guest. Sitting with him, Steve, over to you.
Oh, Zee, thank you so much, my friend. I truly do appreciate that and you. Yeah, I am here in studio with Muhammad Imam, Senior Director of Product Management at Cisco. It's always such a pleasure to sit down with you. We get these opportunities a couple of times a year here at Cisco Live, right? Thank you and welcome in.
Thanks for having me.
All right. Let's go right to the heart of it. Let's go to the C9000 series launch, right? We heard about it at the keynote. It's been a big talk here across the floor. I want to go into some finer detail around that announcement. What makes this 9000 generation different than our prior Catalyst technology that we've talked about so many times?
Yeah. So we just introduced the Cisco C9000 series. Within this series, we are starting with two devices, Cisco 9350 and the Cisco 9610. 9350 is, think about it as the next generation of Catalyst 9300 that we had. It comes with 1.6 Tbps stacking, 90 W on every single port, highest end offer is stackable switches. This is going to future-proof your network for the next decade.
Amazing.
The Cisco 9610, this is our modular core chassis. It's a big chassis. If you remember the Catalyst 6500 back in the days or Catalyst 6800, this is the next generation of that. If you have been waiting for that, remember, LDOS is coming very soon in October. This is the chassis that you want to look at. This is a 10-slot chassis, eight slots for line cards, two slots for supervisors. The box is capable of 51 Tbps .
Wow.
That's a lot of bandwidth with the centralized architecture that we have for the campus switching networks.
The timing is so vital and important as we talk about the speed of application of onboarding AI and capabilities across our organizations, across the network, throughout the data center, and so forth. The need for that additional bandwidth, I think, really has never been stronger. We've got to be able to support people. It's not just a matter of saying, here's what's available to you. We've got to say, and you've got the power to actually deploy it, right?
Absolutely. You need a lot of, there's so much demand in the networks these days. The world is not the same as we had in 2017 when we launched the Catalyst 9000. In the last two years, especially with the evolution of AI, the rapid evolution of AI, we are seeing various different types of traffic in our networks. First of all, the traffic is a lot more.
Yes.
We have data to showcase that. There are hyperscaler backbones that are talking about 30% more traffic in their backbone. A lot of that is coming actually from our networks. The more interesting thing that's happening there is the traffic patterns are changing. Typically, traffic is downwards coming from the data center, the cloud, to the devices. Now, what we are seeing because of the AI applications in our pockets here with the different phones and robotics and so on, there's a lot of information that's actually going up to the models. The AI models that are sitting somewhere in the data center, a lot of the traffic is going up. This is a different traffic, different direction of traffic. We have to make sure that our networks are ready for that. The other thing is this traffic has new patterns. It's bursty traffic.
It is very latency-sensitive traffic. We have to make sure that our networks are ready for this new type of traffic. With this new introduction that we have had with Cisco 9000, we are making sure that we are giving our customers a device and a hardware that is ready to address the AI era.
Confidence, instilling confidence, instilling trust, exactly what Oliver was talking about this morning, right? Muhammad, how are you directly addressing the AI boom in enterprise networks? As we roll out these new technologies, how do we keep that constant pressure, that speed of adoption of AI in our design mind? How do we design for it? Almost, again, future-proof it, right?
Yeah. It's absolutely future-proofing. As I said, part of it is based on what is happening in the AI world. There are two parts. One is the network needs to be ready for AI. We are doing some of the things that I just mentioned for that. The other part is how we are using AI to make our customer's life easier and how we can consume it. That's where AI operations come into picture. We saw some of the new announcements with AI Canvas and the new models. These are some of the innovations that we are bringing across the board. Part of our portfolio with Meraki Dashboard, we are bringing AI Assistant that you can just talk to, to the agent, the AI Assistant, and ask him which AP is down, why it is down.
Instead of going through all the logs that you had to go through, you can just get the answer right away.
Absolutely. Absolutely. This is such exciting technology. Muhammad, I always appreciate these opportunities for us to sit down together. Please keep bringing these fantastic stories to us, sharing it with us here on the Cisco TV space. Congratulations on the great work you're doing and the success this week. Always a pleasure.
Thank you.
Thanks, Muhammad. Z, let's send it back to you.
Thank you, Steve. Muhammad, I tell you two words, future-proof. This goes way beyond those vacuum-sealed bags. I have another guest that I want to bring before you over to Michelle. Michelle, who do you have with you today?
Hi, guys. I'm back on the show floor. I'm with my friend Dan Backman, a Distinguished TME. Dan, what do you do with Hyperfabric? Walk us through this.
I'm on the Hyperfabric product team. I'm responsible for the technical features and functionality of Hyperfabric and Hyperfabric for AI. We are super excited to tell you more about it.
All right. Before we get into the demo, I need to ask, what is Nexus Hyperfabric AI before we dive in?
Absolutely. Hyperfabric for AI is an AI cluster management solution. It is built on top of the same architecture that delivers Hyperfabric data center switching solutions. We use the same controller, the same infrastructure, and the same workflow to continuously design, run, and then build out architectures for data centers. We have expanded this to include AI functionality. This is actually the full AI cluster. It starts with the network, which is really the core of any AI solution. It extends out to GPU servers. We have UCS GPU servers. These are eight-way HGX-class servers with built-in NVLink connections. We also include storage as part of this, as well as management connectivity. It is all these components in one that you require to stand up an AI solution, all in one based on an intuitive controller-based interface.
That's excellent. OK, so how does this help our customers? Break it down for me.
It's pretty simple. Building an AI cluster for any one of our customers is probably the single most complex network and infrastructure build they've ever done in their lives. Even a single 96-node CPU or GPU cluster, rather, can have 500-1,000 different network links. They all have to be plugged into exactly the right place. There are different networks that you plug into. Part of what makes the AI cluster actually work is very careful control of the communications. We have to actually match that up with the collective layer that NVIDIA is building inside their orchestration architecture so that we can actually guarantee seamless data throughput for all these GPUs. An entire GPU server can cost literally hundreds of thousands of dollars. Every second those GPUs are not actually crunching data, your jobs are taking longer, and you're losing money.
We need to make sure that we are constantly feeding these GPUs with data. They never stall. There's never a bottleneck in the network, which is why these are some of the most high-performance networks you've ever seen.
That's excellent. That message just landed for me. OK, last question before we get into the demo. How is this different than what customers have seen before?
It's completely different. One of the challenges of Network and Infrastructure in many cases is you can design it. You can build it. Most of this is by hand. If you start thinking about those thousands of connections I talked about, if you're building a typical cluster, somebody built a spreadsheet that says this port goes to this port with this kind of transceiver, this cruiser of the cable goes. Then somebody has to plug it in, and then you have to configure it. Multiply that by thousands, and you can start to realize just how long it takes to deploy one of these. The most important thing about an AI cluster is these are multi-million dollar investments, usually depreciated over three years.
If it takes your IT team six months to get that cluster up and running, you've already lost one sixth of the value that your company has already invested in that really expensive GPU cluster. The value here is we take that immense complexity, and we help you deploy it very quickly and accurately. Once it's running, we give you the visibility so you can actually troubleshoot and run this cluster to maximize the investment your company has in AI.
Each one of these conversations, I learn more and more and get more and more excited about what we hear at Cisco Live. Dan, I will let you take it away with the demo.
Let's take a look at what Hyperfabric for AI looks like. You log in. This is actually the same controller that powers our Hyperfabric on-prem data center solution. The first thing you're going to do is actually start from a template. These are highly complex architectures, and they are designed and purpose-built to certain sizes. These are actually NVIDIA reference architecture. They're validated, so we work with NVIDIA reference architectures. We're going to go ahead and collect and select one of these. As you'll see, it includes not only network components. We actually support multiple networks, one for back-end and intra-GPU connectivity, others for front-end and storage. Each one of these components is actually built out as part of this blueprint. Now, what's different about Hyperfabric is everything is defined in the blueprint.
Instead of discovering on the ground and managing it, everything starts top-down from this blueprint. We actually compare what we see on the ground to that. Now, your first step when you're deploying this is this template is actually building out all of the cabling information you need. At high speed, especially 400 and 800 Gbps, these optics get incredibly complicated. You actually have to understand a lot of things about how these work through structured cabling. Likewise, there are so many components required to build these that our controller will actually assemble a bill of materials for you. We can even generate an estimate directly with CCO. Why do we do this? It's incredibly complex, and it has to get done right the first time. We're eliminating many errors in this.
Once you actually have this up and running, when you get to day one, actual deployment, somebody's got to plug in those thousand cables. We actually have an interface your techs on site can bring up on their phone, and we can guide them through literally every installation. Every time they plug in a cable, we're validating, did it work? If it didn't work, we're telling them how to fix it. It's a revolutionary new way of not only deploying a data center cluster, but it's invaluable to do in an AI cluster at this level of complexity. We built all of this workflow directly into the controller. This is unique to Hyperfabric. From there, when you're up and running, there's also extensive telemetry into the cluster. Here, we're drilling into one of these 800 Gbps switches.
This is a 64-byte 800 Gbps switch based on our new G200 ASIC set. Here, we're uplifting extensive telemetry. In a cluster this big with network optics this hot, sometimes things happen. You have to keep on monitoring these. The old days of the mainframe, they actually had people swapping out tubes. Sometimes in a big AI cluster, there's people doing that just with transceivers. The ability to monitor this in real time, provide guided feedback to those hands on the ground so they can actually go ahead and automatically repair any issues, that's how we keep the cluster up and running. This is how customers keep on realizing that really expensive investment in the GPUs. Likewise, we also help understand the actual environmental cases here.
We're looking at every piece of information we can get, not only from the switches, but the servers themselves, so we can diagnose any issues with connectivity that could actually lead to interruptions in these AI jobs. This is all included inside that same controller. Also, because this is a cloud-based experience, different people can collaborate on this. It's not just one controller that you log into on-prem. As a customer, you can work with a partner, or you can work with your Cisco account team, not only to spec this out, get it right the first time, but also for help in ongoing monitoring. On top of that, we've also built in extensive in-band monitoring inside the cluster. We can actually understand if we're starting to see network issues. In the end, if you're running an AI cluster, you don't actually have much to look at.
It's a really tough thing as an IT person. Most of this is figuring out where the problem isn't. Because we're providing this level of visibility and this level of troubleshooting and also in-band monitoring, we can very quickly point out any potential issues so you can fix them immediately. And that's a Hyperfabric for AI experience.
Dan, that was incredible. I learned so much. Z, back to you in the control room.
Hold on to your seats, folks. We have DP Venkatesh. He is our Chief Growth Officer here at Cisco. Two brilliant minds are in the room. DP, over to you for a conversation.
All right. Thank you. Here we are in the third day of Cisco Live. A lot of energy, a lot of excitement, lots of announcements as well. Binoy, you saw all the announcements we made. We're very excited about the Cisco Infosys partnership. As you know, it's a 25-year partnership, which is quite a record. When you look at our partnership, what's unique about it is it's both a product, technical, as well as a strategic relationship where not only do we work on products and solutions together, we also go to market together. What do you feel about the 25-year relationship?
Oh, very, very honored and very happy at this event of 25-year solutionship, celebrating together. Actually, today when I was meeting the Cisco teams and Infosys teams, I was telling them this event is like CIS at CIS. I'm happy to represent. CIS is Cloud Infrastructure Security Services at Infosys. We are here at CIS, Cisco Infosys in San Diego. We have been laughing about that. Overall, this has been tremendous, the type of transformations that we have done together, not just as providers to Cisco itself, but then together. When we have worked at clients in the last five years, I would say the momentum really picked up a lot. For example, if we look at the major vehicle, automobile, and trucks manufacturer, we look at one of the largest telcos right here, a heavy engineering consulting firm.
The type of transformation that we are doing for them together, that's amazing. That is the type of outcomes that we together need to drive towards and make sure that we are achieving that journey.
Yeah, absolutely. When you talk about transformation, I think what we both are seeing, and you saw in the show today, is the transformation on AI that's taking place in the market. You saw that Cisco is leaning on our three pillars, which are AI-ready data centers, future-proofing the workspace, and then finally, digital resilience. Now, all three are areas that you play in as well. Is that in aligned with Infosys and your strategy?
Actually, very interestingly, in both keynotes yesterday and today, when I was sitting there and I was listening to Jeetu and the team there, it's almost like you can take my service footprints and say we are exactly in those three buckets, right, of how we are partnering together, bringing the next generation AI-led hybrid cloud journeys, future-proof workplaces for empowering the next level of experience, and creating digital resilience for all workloads, workplace, and the work itself. All of this together, we want to work with Cisco. We actually have two go-to-market platforms. One we call overall as an Infosys Cobalt brand, under which we bring partner reference architectures, repeatable solutions together. That is all in the cloud world. For cloud, it's Infosys Cobalt. The second now, in the AI era, we have it like Infosys Topaz.
Under the Cobalt and Topaz umbrellas, we are bringing these reference architectures for the type of use cases that I earlier spoke about on how we can make it repeatable and deliver the right experience and right outcome for our clients.
That's awesome. Between Cobalt and Topaz, you've got both parts of the journey covered. I think what's unique about the way we go to market together is we are looking at the entire journey, whether it's infrastructure, across networking and security and collaboration. Finally, with observability, there's an end-to-end value that we provide to enterprises. What makes both our companies unique is we're not just providing solutions and business outcomes, which enterprises want, but the real impact we have is our companies are viewed as trusted partners or almost as a guide as people go through transformation in the marketplace. Would you agree that's the way the market views us?
Absolutely. That's a privilege and a responsibility that both of us have. I think Cisco has come up with a very integrated and simplified story. I can think together we are bringing more like integrated, interoperable, and insight-led, AI-led journeys that we are creating for our customers. Really looking forward to do much more of this in the market together.
Awesome, Binoy. Thank you so much for the past 25 years. What I'm really looking forward to is the next 25 years of this partnership. As the world changes and AI changes the world for the better, the impact we will have on enterprises over the next 25 years will be better than what it has been in the last 25. Thank you for that. With that, I want to thank all of you. Enjoy the rest of Cisco Live and have a great day.
Thank you.
Thank you so much. Now, let's see what my pal Robb is doing. Robb, what are you playing with out there on the showroom floor?
What am I playing with? What am I not playing with, I think, is a better question. Here's the thing. When I first ventured out here, I've been trying to figure out why all this entire week I've been trying to figure out why aren't I able to regurgitate more stuff more quickly. As I'm going through the IoT area, I begin to realize it's because there's so much just in this one area alone. You have got so much to do. You can tell by the crowds that are here. Don't look at the guitar activation. It's probably the coolest thing they've got going on here. That's just the tip of the iceberg. If you look over here, they've got the URWB, Ultra Reliable Wireless Backhaul, which I think is on the backside of that thing there. Of course, the guitar demo.
They've had as many as 19 new industrial IoT switches. We're just going to work our way through here. Nope, you guys are fine. Thank you so much. I mean, look at all the hardware. This is why IoT is always my favorite place. If you can, we'll just squeeze in there and take a look at a few things. That's all right. I don't want to interrupt any sales going on. This, again, 19 new. Plus, when it comes to the Ultra Reliable Wireless Backhaul, that is now integrated with the new Wi-Fi 7. I think it's a 6E. I forget all the nomenclature. They're better integrated. That's good because that's the type of low-latency type of stuff we need. Fiber-like backhaul, if you will. We'll come back to this in just a minute by virtue of a video.
I just want to point out real quick. She got campus and branch switching and stuff over here. I want to introduce you to two people that were so friendly. Hello again.
Hello.
How are you? Tell me, let's see your name again. This is funny. Whitney Sales. It's not Whitney in Sales. It's Mrs. Sales.
Yeah.
Yeah, absolutely. In just a couple of words, just tell me. I love the adaptive robotics. What you were saying, if I were to attempt to regurgitate it, Cisco Networking. Here, you guys have been using AI in a very controlled manner to be able to define the robot actually learns how to be more careful about the products it's actually working with. You are in some very unique spaces. I do not know. Did I get any of that right?
Yes. We are adapted for robotic control for dynamic applications, dynamic and dexterous, where there are a lot of edge cases that you would typically have to collect new data every single time there's a change in the environment or to the task. In our case, our models can actually adapt in real time to that change.
That's not normal for AI. Normally, we think about training is a thing. And then you move into inference where you actually use it. They're talking about AI that's learning on a continual basis, literally every single product it touches. They've got many real-world examples. This is just one example of many very cool partner integrations we have. Of course, we're driving it from the IoT switches and such like this. I'm being told I need to move to the guitar area. Julie, I'm going to come back around to your side. You can swing around wide. As we come around here, this is so cool. I'm one of those guys. My video back wall, when I do videos in my home studio, I have some guitars up. Not because I'm a great player. Hard to say I'm even a player. I just like guitars.
I like musicians. I love music. That's not really the point here, is they wanted something that looked really cool but could be a demonstration of how you can actually check for all these things using Splunk dashboards, integrated technologies from the AI in a box using the UCS and the things that we've rolled out here. It's just fascinating. I love the IoT group. Steve actually did a full walkthrough talking us through this demonstration. I don't want to spoil anything he may have said there. Why don't we just go ahead and take a look at that video now?
All right. I'm telling you, you guys have got to get in here and check this out. Factory of the Future, 19 new technologies from Cisco. One of my favorite things that we're looking at here on the show floor of Cisco Live 2025. This is our AI-driven industrial manufacturing demo. Awesome story from Cisco Industrial IT in partnership with Denali and Minds Mechanical. They helped us build this out. It's so amazing. What you're looking at is anomaly and defect detection in an industrial environment. I want to walk you through what you're actually seeing here, right? We're using the convergence of IT and OT, plus a lot of new technologies, like I said, 19 new technologies purpose-built for your changing environments across your factory floor. First here in station number one, we are using AI machine vision inspection to find flaws in the guitar here, right?
We've got three Cisco business units working in partnership. Industrial, Splunk, and Compute are working across this space, right? The AI is looking for scratches and dents anywhere within the camera area. It's going to discover defects in the guitar. We move back over here on this screen. We can see the defects. You know that AI is going to add to bandwidth traffic, right? That's going to create a lot more usage across your industrial network. Like I said, Cisco is adding new devices right down here that accommodate that higher bandwidth within your industrial networks right down here in the box, right? Follow me. We'll check this out a little bit later. Back over here on this side on station two, this is so cool. I love this. I want to introduce you to Axel. This is Axel, right? He's a robot.
What he's doing is tuning the guitar based on frequencies that come out of this unit. Here is how it works. He lifts the guitar. He is going to steady it against a pickup. He is then going to pluck just one of the strings at a time, right? The signal from that pickup is going to come right over here into this booster pedal, the red box that you see. That goes into the headphone amplifier on this side. It then goes into the oscilloscope back over here. The signal is then going to go to the virtual PLC inside the server, right? That PLC is going to tell Axel how to tune that one string based on the frequency that that string puts out. By the way, this is just to show what we can do. This robot, Axel, can be programmed to do anything, right?
Our devices are expanding all the time, all the different potentials and possibilities within your industrial networks. One more station. Come on over here. Station number three. Now we are talking Splunk, right? We've got this screen. This is our dashboard for the complete machine, the total system. On one side of the screen, we can see production. On the other side, this is what is happening across the infrastructure. We've got visuals of the switches as well. Like I said, this is an incredible demo, great capabilities. In fact, I'm going to come right over here. Come on back with me. Cisco Industrial IoT. Get in here. I know we're making you like chase the cameras around. Thank you, my friend. You guys put together such a cool setup, you and James and Richard and the whole team here. One key takeaway. People come by.
They want to check this out. They want to see what's going on with Cisco and AI industrial-driven manufacturing. What are we doing? What is the key takeaway?
Key takeaway is Cisco Industrial IoT enables AI in industrial networks that are critical for the lines of business that we support our customers with.
Love that. We can customize this to whoever walks in the door and tell them exactly what we can do for them with this new technology, right?
Absolutely.
Factory of the Future, 19 new technologies to check out. Justin, thank you. Congratulations again to you and the team. Like I said, get in here. You got to check this out.
Folks, that was amazing. Axel, the guitar. I tell you, folks, AI is doing so many things in this new industrial manufacturing industry. I tell you, you need to be here to see this. So many activations are on our showcase floor. You want to be here to see this on next year. You need to be trying to figure it out. Get your life together. You need to be here for Cisco Live 2026. I'm just setting you up. Coming up next, there's more. We're about to go to center stage. So much information that we are going to be consuming and unloading on you. So many experiences that you need to be here to experience for yourself. Let's go to center stage right now. Take a look.
Good afternoon, everybody. Thank you for joining us. Hope everyone had a great lunch. We are going to dive into a topic that I am very, very passionate about and something I have spent the better part of the last two years of my life talking about, which is AI, AI infrastructure, and really what we need to be considering when we think about what success looks like. I love to do this to my audience. I love to give pop quizzes. Real quickly, if I say the date November 30th, 2022, and shout it out when you think about what it is, why is that date significant? ChatGPT. That is when ChatGPT became publicly available. That was right just in time for Christmas. By Valentine's Day, the platform had reached 100 million users.
In less than 60 days, we saw something that was unheard of outside of research and academic communities to being mainstream. We haven't really looked back since. We think about it because it has shaped so much of the conversation around what we do and around IT. We do the math. That was less than 31 months ago, less than three years, and how it's fundamentally shifted things. As an IT practitioner, we're now dealing with, oh, use Webex. Sorry. We're now dealing with a completely different set of challenges. First and foremost, everybody overnight became an AI expert. Our customers' executives are getting pressure from their boards, in many ways, just to do something because they see the capabilities that this is going to bring to their business. We're seeing, as I mentioned earlier, everyone has become an AI expert.
We have lines of business that are going out and using tools that are not IT approved. One of the customers I'm very close to is a large research hospital. Their Head of Data Science told me that the way they find out about new AI deployments in their organization is when they get calls from the help desk complaining about the noise and heat that servers that have been stuck under people's desks are creating. There is a governance challenge that we're seeing. Security is in and of itself a conversation that we could have for hours. Just think about this. There have been several high-profile challenges here. Anytime we interact with one of those public models, ChatGPT, Gemini, we're training that model on the fly. We have seen instances where proprietary data and confidential data has been released on the internet.
We have to think about that as a security concern, but also how do we make sure the models themselves have not been corrupted from when we bring them into our applications. We'll talk a little bit more about that. Then deployments. Where are we going to put it? How are we going to put it? Is it going to be cloud? Is it going to be on-prem? Will it be hybrid? Is our workforce ready for the challenges and the capabilities of Generative AI? Actually, I want to talk a little bit more about that. We recently released our second AI readiness index. The numbers were pretty telling. Again, as we talked about early, 98% are feeling the urgency.
Half the customers we talked to are dedicating between 10% and 30% of their IT infrastructure or their IT infrastructure spend towards Generative AI projects. Yet when we did the math and we crunched the numbers, we found that only 12% of organizations were truly ready to start deploying Generative AI in their environments, either in the cloud or on-prem, because of all those challenges that we just talked about. That is why we're seeing a real challenge between the hype and the promise that Generative AI is going to bring and what organizations are seeing in real-time deployment. In fact, 88% of the POCs that we have observed have either failed or they have not lived up to the financials to go from pilot into mainstream deployment. Why is that? Because it's fundamentally challenging how we build infrastructure. A lot of organizations aren't really ready.
One of the areas that we're acutely focused on here at Cisco, for probably obvious reasons, is the impact that these are going to place on the network itself. Whether it's training or inference, we've learned through observing either the hyperscalers or large enterprises that the network truly matters here. Even losing one or a few packets in one of these large training jobs, or you might have heard of the YOLO jobs, you only live once jobs, can corrupt the entire workload. The importance of the network cannot be understated. We talk a lot about GPUs, but really the network is the critical linchpin in how customers are receiving value and able to drive this. Second is security. Just a quick show of hands. Who here has heard of a website called Hugging Face? All right.
If you have it, you need to research it. Here's why. Okay, let me ask this other question. Who here has heard of GitHub? A lot more hands went up. Think of Hugging Face as the GitHub for Generative AI models. There are literally thousands of models being released. They're based off of largely the foundational models that we've come to know and are now household names like GPT, Gemini, Bard, what have you. The challenge is that these have not gone through the rigor and scrutiny that those large foundational models are. We have seen several instances of those models coming into organizations already corrupted with false data. This is a real challenge outside of the traditional security paradigm, making sure that these models are appropriate for use and that there's not misuse built into the model from day one. Then visibility.
Let me ask this question. Are GPUs cheap? No. Jensen likes it that way. Trust me. The last thing that you want is idle capacity with your GPUs. You want them as fully utilized as possible. Being able to understand and get visibility into the interaction between those GPUs, into those training and inference jobs, is critical because we want to optimize that as much as possible. All of these things I have talked about so far, these are what and how conversations. Why are organizations even doing this? This is something I often talk to my own team about, no one wants to deploy AI, just like no one really wants to build networks. They have to build networks to support their business and support their applications.
The why behind these Generative AI projects and why we're getting the pressure from the CEOs and the boards to do something can actually be very well summarized by this slide, is because we think that we can take the massive amounts of organizational data and make very, very informed business decisions. Again, one of them, one of the actually most common is this first one around detection and prediction, which predates Generative AI. Good old-fashioned SIM and SOAR and how organizations are looking at incidents and correlating data across thousands and millions of data points. We get into specialized research. This is where some real breakthroughs are occurring in the industry every day, where we're getting into personalized medicine, genomic mapping based on the ability of Generative AI to look at individual patients and correlate that data.
One area that we're seeing a lot of investment and a lot of conversation is in data sovereignty. In fact, we just made some very large announcements with the Kingdom of Saudi Arabia around sovereign AI deployments because organizations and countries see this as an opportunity to provide a next level of service for their residents. The one I'm going to close on is one that is near and dear to my heart, is infrastructure optimization. How do we make sure, as I mentioned earlier, around getting full utilization of those Generative AI clusters, but how are we leveraging Generative AI to optimize the services that we're providing from Network and Infrastructure? I can't think of anyone better to talk about that specific topic.
More broadly, what we're doing within Cisco CX to drive Generative AI success for all of you, than my very good friend and colleague, Carlos Pereira, who is a Cisco Fellow and our Chief Architect in Cisco CX. Carlos, come on stage.
Thank you. Thanks, man. Thank you. Thank you very much. Thank you, sir. No problem. Welcome, everybody. By now, you look like a PhD on problem statements and the potential solutions. The question is, your mind is, how can Cisco Customer Experience help you? It's very simple, the question, but the answer is the one I want to articulate to you a little bit more. There are three areas that Cisco CX, Customer Experience, CX for short, can help you. One is help you define AI for what. A lot of customers come to us like, "I want to use AI. I want AI. I have this ChatGPT and the new ones, but really, beyond a chatbot, what's my use case that actually brings business impact?" After that, it's an area that we can help. We're going to have a dedicated session for this also tomorrow.
We land all of those needs to run somewhere. It needs to be secure. Let's go one by one. As it relates to define the use case, we have inside CX what we call an AI adoption framework. It's a methodology that allows you to actually start for the identification of the potential use cases. That comes from the business side, sometimes from the IT sides, usually a combination of both, that actually how can AI help the teams and the process to get better. It comes from the use case prioritization all the way to define the design of the solution that we apply for that. You need to validate. The validation is not necessarily the models, like Eric was mentioning, but validation has to do also with the data sets.
If I have a problem statement and I have a use case and I try to have AI to assist and augment, there is a lot of data that goes there. Not necessarily all the data is unstructured data. Not everything is a chatbot that is dealing with a historical text. There is a lot of SQL stuff somewhere that is someone who needs to denormalize, actually defragment, and stuff like that because garbage in, garbage out on the AI models, right? You do the validation, potentially an MVP or a proof of concept. From there, you go build on the infrastructure, deploy, adopt at scale because you may target a specific audience or you may go to the public. Obviously you monitor and operate.
This slide is just a reference and just a cookie for tomorrow because tomorrow there is a session that I'm going to cover this with. The former CEO of Deeper Insights, a company that we acquired that has been doing this for 10 years, AI, not only LLMs or SLMs, large machine models or short language models. It's also for computer vision models, for predictive models, for drift models, everything that relates with AI. This is a reference for you. I'm not going to drive this today, but I'm going to share with you some experience on what we've learned interacting with our customers and ourselves with our own CX internally. The first thing is define the use case first. Define the success metrics for those use cases first. There's a lot of shiny objects on AI.
Believe me, by the end of the conference, there's going to be a handful of more new tools out there that's going to look cool. There's a lot of pseudo experts on AI on LinkedIn as well. Talk about RAGs, supervised fine-tuning, few shots, low shots, reasoning. All of that is good. It's important. If you don't have a purpose for that, you're not going to justify how you're going to pay for those GPUs or the infrastructure. That's first. The second one, as I mentioned before, as you go to enterprise-class use cases, you most likely will need to augment your model with insights from data. That data most likely is on system of records. I'm going to tell you, LLM is a three-letter acronym. SQL is another three-letter acronym. They don't date. They go on a long wheel. They don't like each other.
You need to deal with a lot of the data that sometimes is not very easy to deal with in order for the model to give accurate and proper results. Think about that. Denormalize the data first and go in that regards. Another point to keep in mind, experimentation and rapid prototyping is very important. The speed as we run enterprise applications in production is not the same speed that your AI team will need to experiment. Why? Because AI is changing much faster than everything else. You need to have a dedicated team to think about that. They're going to fail fast and fail more. Why? Because the unpredictable nature of AI as a technology. While the enterprise use case is your business guy is going to say, "I don't care if it fails. If it's new, I still can lose the money here.
I still need to be competitive and stuff like that. Last but not least, avoid hype. AI demos for a Twitter post is very easy to do. Making applications that run in production at scale is not that simple. We are here to help you out. As we come from the use case identification, we need to see how we can help you on the AI infrastructure and the security that comes with that. For that, we need to help you from the lens of what we call a life cycle. The life cycle is started with the planning and the design, as I mentioned before, and the implementation, which are typically on-demand kind of activities. You do it once. You need to refresh, and you need to expand. You do that.
Everything that's on the other side of that slide, migrate, operate, and optimize, your left-hand side is ongoing. Operations on a daily basis are the kind of activities that you need to do. Before we help you or I tell you how we can help with automation and AI through the life cycle, let me define what we mean by life cycle. Life cycle, let's think about life cycle as something that has to do with time. The first thing that we can help you is as you want to accomplish something, we define the use case before. You have an intent on your mind. Here's my desired state. I want to get somewhere. CX helps you with the plan with that initially. When you go to the design, not necessarily everything that we intended might be possible because the technology may have shortcomings.
Your budget may have a shortcoming, or maybe the technology is not at the level that we want to be, or you need to phase it out. You have this high-level design with Cisco CX helps you out. More than anything, the low-level design. That is where you leverage a lot of the experience that we have for years on stuff that works on other customers or failures that we saw happening or some of the things that need to be taken into consideration. By the time you have this, you start your implementation day. That is the first day of implementation. You go and do the first deploy. Over time, you are on operations. Guess what happens? You have an intended plan. You have the design. As operation goes by, not necessarily you adopted everything you said you would.
Not necessarily the scale that you planned might be bigger or smaller than you anticipated. Or we may have some new use cases that come along. Because we work on CX close to you, close to your teams, we can predict where the expected state would be. Why? Because let's say you want to implement for 50 customers, and as of a certain, you have 200. Obviously, you can predict that you need to 4x one infrastructure. That was an easy math, but you get the point that I'm trying to make. Let's say you need to get on a change. When you get on a change, then you start to face reality. Why? Because there is typically a drift between what's the intended state and the reality. Why there is a drift?
Because we may have introduced a bug or a defect, or you may fat finger something, or you may actually not have done a plan accordingly, or there might be an industry security vulnerability that you need to patch, and you have not patched the complete landscape. There are multiple reasons. The first thing we can help you with AI is actually get services that allow you to minimize that green triangle. How can I bring it back with an adoption AI journey? What is the ideal adoption journey for that particular intent that you had versus what is actually running? You have not adopted all the functionality that you have bought. How can you help you do that? There are a lot of situations. Minimize that initial green triangle.
The second part, which is very important, is how can I minimize the second one, which is through proactive services, minimize that drift. Use services as code, as a mechanism of automation and insights to identify that area to try to bring this back. Let's say, "Hey, you are exposed for that security vulnerability in half of your footprint. Let's automate that. Let's get out of this new release. And if you get to this new release, you're going to have this or that impact." All of this is proactive ahead of you so you can actually work together on that. That relates to the AI. Jesus Christ. That has to do with AI use cases. Hopefully, everybody's okay back then. We'll check later. And he's coughing, so it's alive at least.
Outside of that, it comes to a point that you can have those two services that CX can help you on the infrastructure, on the security for the AI use case. It comes to a point that you may want to expand. You want to expand to a new desired state. You made an acquisition for a new company. You hire new people. You want to go and pursue a new area on the market that your company wants to get there. You come from a brownfield now. You're not a greenfield starting from scratch. You go to that place. You start from the same place. Cisco CX can help you with the visibility, a comprehensive visibility that has AI assurance and AI, not assurance, AI assistant to make you assured that you can get to that point.
At the same time, bring in product support and bring troubleshooting into this. Talking about troubleshooting, let's assume that before you reach your future on time number four, on T3, a vulnerability kicks you and it's discovered in the industry as what you're vulnerable. We can actually help you to predict what's the new expected state and what's the potential disruption impact that it's going to have on AI infrastructure for you. With security assessment that are AI-assisted, we can help you on that. I'm going to give you a very quick example on a demo that shows the middle one in product support, how we brought AI together from the tech to Cisco Technical Assistance Center into the product to avoid the swivel shared operation challenge when it comes to AI. Let's roll the video, please.
In this demo, we're going to show you the integration between the Cisco Support Assistant from TAC and the Unified AI Assistant for Security, which allows all the in-context information from TAC to proactively be rendered inside the AI Assistant in the product. The SecOps operation avoids the swivel chair effect and has a unified experience when he needs troubleshooting or any advice from TAC. If you see what's going on, the in-product AI Assistant is now open through the click on the field notice notification that was proactively generated from TAC. The SecOps person is now trying to find more details on what this field notice is all about, which the AI Assistant provides at the hands of the operator. Additional information is needed. You want to know what versions that field notice applies to you.
The AI Assistant is contextualized, and you show exactly in your environment. Moreover, the user is trying to find out how many TAC cases may exist as it relates to that particular field notice. With all this in-context information on their hands, the SecOps operation sees that, "Hey, I have some TAC cases that were open, and they are still pending in engineering. Assign it to that. Can you please provide me a solution?" As you can see, this is a SecOps in-product interacting with the TAC without needing to open a case, go into the TAC AI Assistant, or have an interface there. It removes the swivel chairs, and now the information from TAC is automatically rendered in the screen of the AI Assistant and the product for the SecOps person.
Now the interaction still goes when the SecOps person is like talking with the TAC Assistant and trying to ask help for a Vicky firewall. The TAC person has no idea what the Vicky firewall is, but in-context on the product does. That correlation goes across. In fact, the SecOps asked to open a case and changed their mind. The case was successfully opened, but through the SecOps interface with the AI Assistant, the person asked the case to actually be closed because he or she was able to implement the workaround by themselves and deploy the needed policy in the firewall through the FMC. The AI Assistant interacts with the TAC and actually closes the case, and everything goes smoothly from an operations perspective.
This is a demo that you can see the reality coming across. Even myself, me stamping a lot of times and messing around with the Assistant, which actually happens. By the way, there is nothing wrong there, just a ladder fall. Everybody's okay in the back. I went there to double-check. Let's keep rolling then. Another area that CX can help you immensely is as you go to automate the infrastructure that I mentioned for AI. Typically, what happens is you try to go for deployment. You configure a device, and you local test, and it may fail. You go ahead, and you regress the test. It may succeed, but it may also fail. You keep going on that flow, and it's very serial in such a way that you get to change on the production, and you finally say, "Guess what? The change happened." Great. Then you remember, "Damn it.
I need to manually create all the documentation of what I just did." And it fails a couple of times, so that's going to be a lot of work. We changed that by creating a data model of all the infrastructure in order to minimize for this 24-48 hours typical process that exists for automation of a lot of infrastructure. We model the infrastructure underneath in such a way that we then, because you have this model, it's independent of infrastructure. You can run tests on a DevOps environment. We put this on a Git DevOps. We get this as part of the test and fully automated. Not only that, we can also automate the documentation. It changes the automation or the process because of automation for hours to eventually minutes.
Now, by doing so, a lot of people can actually benefit because we actually abstracted from the infrastructure itself how you do tests, how you do validation, and you automate. You use the open-source models that exist out there, like the Terraformers and the Ansibles of the world. We just had the schema for the use case that you need to implement. We went one step further. We went and also now use GenAI to actually generate the code for that potential implementation. Beyond that, which is more interesting, to generate scenarios for the testing. You have hypothesis-based scenarios that you test automatically against, which makes it more rich than someone could build across because now you can test automatically, but you can test across multiple other combinations that minimize the risk that may exist for you.
Let me give an example on how we can get there. That was also thought with all the people and operators in mind because let's imagine you have a new use case of AI. Great. Your boss wants you to put this in production. Great. You don't know all the AI yet. And then you go to infrastructure new to your job. You're giving something like Cisco ACI, and then your question is, "I'm new to this config. Can you help me out?" Before you go to the whole documentation, all that green, either greenfield or brownfield that I said before, you can ask the Assistant, and it's going to summarize and give you an explanation of what the configuration is and what it does. Tomorrow, we have a session that's going to be 30 minutes specific on that topic with demos that you can also see.
Another use case that is typical is as you understand what a configuration does, you had your intent. Remember? The configuration is here. I want to do something there. I want, let's say, to take that configuration and add a subnet. It's a silly example, but it's just to simplify what I'm talking about. You have an environment that already exists I want to change. Instead of you going through the documentation, ask the Assistant, and the Assistant is not going to go to the documentation. It's going to go to your intent that I mentioned before. It's going to go through official documentation and the data model for the automation and go and bring it back for the test scenarios for you. It's going to tell you what you need to do and what you need to implement.
In the end, it's going to even say, "Hey, dude, you remember that you need to put an IP address on the damn VLAN? Otherwise, it doesn't route." Kind of basics, right? The last but not least is the basic scenario of an error or troubleshooting. Something goes up and shows you an error. You have no idea how to decrypt that, let alone to understand. You copy-paste, put it there. It's going to analyze for you and say, "Hey, I see there is an error. Let me check that configuration. This is a corrected configuration that should be implemented on that particular scenario for that intent." You have two options. Either you automate all the way through, or you can open a ticket with ServiceNow and follow your procedures for change management. That's pretty much what we have done here so far.
Last but not least is the security. I want to spend a couple of moments because that has a little bit of concept that comes here. I'm assuming that the vast majority of you, sometime in your life, you have programmed an application that says, "Hello, world." Right? Even my beloved friend Adele and Jackie did from CX that are over here. If you hit the "Hello, world" 100 times, what are you going to head 100 times back? "Hello, world." Right? On the AI application, it's a little bit different. You're no longer on the left having infrastructure data, application, and user consuming that. You inject now a variable that's called the AI model. Large language model, small language model, the new one, agentic, it doesn't matter. That layer is not deterministic, which makes your whole stack non-deterministic. Right?
How do you manage with that? How do you deal with that? There is an aspect of risk vectors that are novel, if you will, from the lens of the application development stack that you need to be considering. That's where, if you think on how people develop applications in general for AI, they take an open-source model that's out there that has been trained on public data, and then they augment that through RAG accessing information that is on a database or even for supervisor fine-tuning, multiple mechanisms to get this sort of production-ready model, especially for GenAI, and they use their interface with this and get an answer back. You have a context that needs to be carried on top of that. It's very high level. It's more complex than there sometimes.
What happens is you have traditional security issues that can happen for any application, like identity stealing or data exfiltration or denial of services, all of that. You have a new category of safety risks and issues that may raise, like hallucinations on the model. Someone tried to influence the prompt or actually the context not being carried properly. There is a misalignment. There's a lot of interesting things that exist in the new universe that makes that unpredicted or non-deterministic nature worse. Cisco has brought to market something called AI Defense that does two things. One, it helps you with that safety and security assessment, and it's the right-hand side of that slide.
It talks with the model providers, has support with all the models out there, the Geminis, the OpenAIs, the Mistrals, the Anthropics of the world, but also connected data sources, as I mentioned before. In the middle, it does application security AI and actually informs you for the notion of shadow AI. Yes. Remember shadow IT back in the days of the cloud? Why is it going to be any different now? It's actually happening. It also touches the left-hand side, your left on this slide, which is basically how I can actually AI security interleave and connect to the enforcement points that you already have. Like you have single sign-on. You have security access control. You have all the base that has the defense for the perimeter defense. All of that is already on your infra. How can you make those two things work together?
Because the majority of the solutions for AI today, they focus on AI only, which is good, but they do not necessarily map to the enterprise security that exists out there. With that said, those are the three areas that I want to bring my big friend Eric back, that CX can be helping on infrastructure and security by the time we identify the use case. Eric, why do not you bring us home?
All right. Thank you so much, Carlos. And everyone, thank you so much for joining us today. I think the message is very clear. AI is transforming all aspects of how we deliver our services, how we build networks, and ultimately what you are being expected to do as an infrastructure practitioner. This is about artificial intelligence and human intelligence coming together to deliver an exceptional experience for you and your organizations.
Our call to action here is simple. Go ahead and jump ahead for me. Right? First and foremost, members of my team, please raise your hand. All right? We came in force. Please reach out to your CX leadership team, the CX team that covers you. We're driving some tremendous innovation across Cisco. Go to the World of Solutions, check out the Solutions Showcase. Look at how we're embedding AI into each of our capabilities. With that, we want to thank you so much for joining us today and enjoy the rest of Cisco Live. Thank you.
Wow. That last session, Eric and Carlos delivered. I tell you, the way they broke down AI integration, we're talking smart, we're talking clear, and honestly, super timely. I'm here with Steve in the studio. Steve, what do you think about that last conversation?
You know, this is kind of an amazing thing. I love to hear people talk AI because a lot of times we do not really understand the depth and the breadth of the story. This is not the kind of thing that we can wrap our head around for our own capability and our own knowledge. When we talk about the power of AI, we start to think about all the different areas that AI is not just tripping into our lives, into our day-to-day, but starting to really alter the nature of our day-to-day, the way we think, the way we work, the way we live humanly, the benefits it is providing to us, right? The things that Cisco is supporting that we just heard about, the way data-driven AI is reshaping full industries.
Not just processing data with unprecedented speed and accuracy is literally predicting trends for us now, right? Providing strategic insights that are redefining industry standards. We look at AI's role in well-being, right? In us. And it's so much more than we think. AI is being used right now to harness better work-life balance, to monitor our stress levels, to promote mindfulness, promote healthier work habits. It's not just technical. AI is becoming a partner with us to foster a better human experience that we talk about. We look at AI models, right? And it's going beyond human intelligence now that we used to talk about. I was hearing it earlier today. AI is being used in advanced research labs right now. It's being used to simulate chemical reactions, create molecular behaviors, and predict them to revolutionize drug discoveries for us.
We look ahead and we start to see how AI will not just assist scientists in groundbreaking ways. It is going to be collaborating with us, helping us create solutions we have not even thought of yet. I will mention one more: capital markets. Right? We talk about finance. Everything from overall operational strategies to creating shifts in the marketplace. We are seeing AI deliver a lot more business-friendly regulatory environments to encourage investment, encourage growth. Imagine a financial landscape where AI-driven analytics is now improving things like risk assessment and compliance reporting and allowing our institutions to be more agile, more precise. So many great, exciting things. That is why I love to look at a center stage session like this one, Z.
Oh, amazing. AI, AI, AI, AI, AI. Let us go to my co-host, Michelle.
She's on the showcase floor with someone that's, I'm telling you, he's going to talk about something amazing. I know he is. Michelle, over to you.
Z, how did you know we are going to talk about something amazing? I'm here with Bhaskar, the SVP of Engineering for Cisco Customer Experience. We're in the CX showcase. I'm going to start out with some questions for you. First, thank you for joining me.
Michelle, glad to be here.
Excellent. Laying a secure foundation for our customers. Why is it critical to lay a secure foundation for networks?
We've been talking a fair bit about digital resilience. Digital resilience is all about running your network securely and efficiently while you're handling a lot of changes. One key part of that is security.
With the sheer number of threat vectors you have today, keeping your network secure from day one is very critical.
We have been getting a lot of very interesting, groundbreaking announcements this week. Can you walk me through one or two?
Oh, it has been, and this is day three. A lot of announcements. There is a lot of exciting stuff. We talked about it on center stage. Liz talked about it a little bit. I am going to give you a sneak peek of one part of it. How do you ensure that your network is hardened and secure from day one? At Cisco, we put out hardening guides. I mean, with a lot of different details. Usually, you will have to pass through all of it, make sure that you are actually applying all of these rules.
What we have done is, with the power of AI, made it pretty simple for you to use. We've made it into a playbook, which is nice and categorized. As long as you can connect your device, you can connect your network elements onto the system. As you can see here, it's nicely categorized into security advisories, management plane, data plane, control plane, and all the supplementary checks. You have a lot of these rules. These are hundreds of rules, and usually, they are time-consuming. What we've done here is made it really easy for you. With one button, you can pick all your security checks. Press the button, the system would automatically run it, collect the configuration, analyze it, and make it easy for you to understand what is going on in the network. As you can see, voilà, it's run it.
You can see that the network is secure from 10 different threat vectors, but it is exposed to another 10. It gives you all the details of what's happening here. This is the best part because we are using the power of AI. It can explain it to you in the context of your network. We also allow you to have to remediate it right from here. With, again, a press of a button, you can ask specific questions about a particular threat vector. You might want to know a little bit more about it. Given that remediation is a comparatively hard problem, we generate the code, which you can edit and make sure that you can push those changes and remediate a particular threat. You're going to see this run a little bit. I mean, this is the code that was generated.
Maybe you want a little bit more detail. With the AI assistant right there, you can ask for it to add comments so that you can understand it better. You execute it right there. It executes it in the context of your network. After it executes it, it does the check all over again and verifies that, hey, now you are safe. With all of this, you get a secure foundation. You can harden your networks from day one. These hardening guides also keep getting updated, or there are new security vulnerabilities that come out every day. With this, you are ready to go at any point in time. On demand, run your security assessment, and there you go.
Bhaskar, thank you so much for your time. Z, I hope you liked that. Now, back to you.
I did, Michelle. Robb, who do you have? Over to you.
Oh my gosh. We're talking secure SD-WAN. A m I pronouncing that correctly?
Yes, Robb.
Tell me your title again. What are you responsible for?
I lead the Outbound Product Management Team for the Enterprise Routing SD-WAN group.
Perfect. And you guys have, just like we've, this is amazing to me because I was complimenting on how there's so many people who have done a complete portfolio refresh, really.
Yes.
I w onder if you could break down not only what you have that's new, but what is it for those of, for anyone that's familiar with what we've been doing, what kind of changes, what's important to understand here?
Yeah, sure, Robb. This is the massive innovation we are driving with launching the new secure router 8000 series product family, where we are converging Catalyst routers and Meraki MX products.
Oh my gosh. T hat's key, right?
Yes. Because that's the surprise, actually.
Now, okay, that's simplification across the board.
With this, you will no longer see any new product with Catalyst routing or Meraki MX routing as the name because we are converging the portfolio, the same hardware to be used for Meraki MX as well as Catalyst to drive the next generation secure WAN to address SD-WAN routing SASE requirements. The best thing about these routers is these routers are built on purpose-built ASIC to address what we call security acceleration, data plane acceleration. Also, some of the high-ends have ML chipsets to provide investment AI ML chipset to address the next generation future AI ML requirements for the customers. What we are looking at is distributed AI Agentic AI requirements, like what Jeetu spoke about during keynote, the Canvas AI.
Definitely, that would be integrated to this solution in the future. We have a strong roadmap.
I like you got unique new hardware as well.
Yes.
Can I just say, I very much love how you guys have gone to the extra trouble of lighting along the edges, the breakaways, because no switch I ever actually worked with looks this pretty, of course. In the rack, it's usually too dusty that it's more dusty.
These lightings have been provided to showcase what we have done in it. Like, these are the first time addressing branch, campus, and data center edge requirements with 100 Gbps ports. This is the first time in enterprise routing SD-WAN portfolio that anyone in the industry is introducing 100 Gbps ports. This router I have, 8500, has an ability to support 200 Gbps ports or even 440 Gbps ports.
Oh, wow. [crosstalk] is close up on here. Can you point out kind of some of the new components?
Yes. This is the first time in the industry anyone is able to support 100 Gbps ports to address edge SD-WAN requirements, aggregation requirements.
Are you seeing us moving more into edge requirements for AI inferencing loads?
Exactly. That is because we are looking to address all distributed AI requirements, which are going to power the future data requirements while also securing the network, while converging the entire portfolio so that customers have consistent unified experience under a single dashboard for the entire enterprise networking portfolio.
That is amazing because it feels like that, it feels like you guys have, it feels like such a big turn that you have made all of a sudden as well.
You said it's $5 billion, something like that, invested in this material change and this type of thing. What do you think is most important? Are there issues that you need customers to kind of mind blocks to overcome to understand real value here? What's important to understand about what you're doing now versus how you've done it before, maybe besides just the management or the integration, although I think that is one of the most important?
A few things which we are doing, which we have never done before. One, these routers would be managed through our full stack management dashboard, which is Meraki dashboard. iOS XC routers can be now managed by Meraki dashboard.
Any customer who has a lean IT where they want a single dashboard to manage their full stack, which is router, switch, wireless, these routers are a perfect fit for that. Also, as I said before, with these routers, any large and medium enterprises where they have dedicated SecOps, DevOps, and NetOps separate personas, these routers are able to address that. These routers would be supported and managed by Cisco SCC, Secure Cloud Controller. [crosstalk] Yeah. SecOps, which have a very dedicated security requirement, can manage these routers, the firewall stack, the security stacks of this router in the exact same fashion they manage firewall appliance or SSE, Cisco Secure Access in the cloud security space. One single dashboard for SecOps users to manage entire security need, including on the WAN side.
Excellent. Just to make sure I'm clear, if I'm an engineer who is looking at deploying these, and I've got knowledge of everything you've been doing previously, what is the learning curve or anything I need to be aware of for making the adjustment? Because it sounds like you're dealing with a lot of familiar things, but I got to mean there's some changes.
Robb, as we have been saying before, our focus has been simplicity, ease of use. Yeah. That means your learning curve reduces drastically. Everything can be done through a very easy-to-use guided workflow-based dashboard, monitoring, orchestration, management. That means your learning curve is very minimal. You no longer need to have those extensive readings, studying CCIE certifications to be able to manage. Definitely, a CCIE expert would be able to manage them with ease.
Any other Nova-ish admin should be able to manage it easily.
Perfect. Agarwal, thank you so much. I appreciate your time. This is amazing. It was kind of hard getting in here because there are so many people crowding in here. Your team has been wonderful to work with.
We have been getting such a positive response on this.
I see. I can see exactly why you would. Thank you for taking the time.
Thanks a lot, Robb.
With that, we will throw back to the studio.
I love it, Robb. I took away some great commentary. He talked about a unified Customer Experience on a single dashboard, reducing clicks. That is going to be something simple and ease of use. I love it. So many special things are going on on the showcase floor. So many things to see and do.
We have coming up next, back on center stage, data center security overload, architecting networks for AI scale. We're going to see how Cisco is building faster, safer, and easier to manage networks by adding security and services directly into the network. We're embedding it into the network. We're starting at the infrastructure using tools like hybrid mesh firewall and Splunk to cut costs. We love a cost savings, no more cutting coupons and boost performance. We want to boost performance. Let's go onto the showroom, back to the center stage, excuse me, and we're going to see what's next. See you there.
Hi everybody. How are you all doing? Thank you for taking the time today. Craig and I and our guest, Neil, are going to go over the next 30 minutes, talk about how fundamentally transforming your enterprise architecture is a key element of the evolution of data centers. Without further ado, as it was just being described in the abstract, enterprises are going through a fundamental reshift. I mean, we are looking at environments that are hyper-distributed because of a distributed data domain where applications are being managed as well as how users are accessing infrastructure. In such an environment, essentially, the operators' resources, they're not expanding. Their management expects them to manage the same kind of infrastructure with this massively expanded attack vector from a security point of view with the same amount of visibility, tools, and resources. How do you do that? Essentially, that problem is what enterprises are grappling with.
The traditional models and molds of building architectures with centralized services domains and a network that caters to that is not going to cut it. Fundamentally, these are the four areas of problems that they're seeing. Now, one of the things that we've done about 14 years ago, Craig, if you remember, when we went on the whole journey of SDN, of Software-Defined Networking, was that we moved from network architectures that were box-by-box architectures with management of those boxes in a single instance, one by one. We moved to a system-based approach with centrally defined policies and distributed enforcement. What we've essentially achieved with that was the ability to go and deploy networks at scale across multiple sites as well. With such an environment, we now are building upon that by bringing in security infused into the network.
We can't think of security as just a bolt-on environment anymore. That's where consolidating those architectures in such a way that your network essentially has security infused in it from day one, the ability to define the centralized policies as well as distributed enforcement are key. Driving the observability of that and ultimately future-proofing that architecture is going to be a key thing. How do we do that? One of the key things Cisco is very good at is starting from the first principles. The first principles start from silicon systems, software, and the operating model that goes around it.
Once you build that four principles, foundational pillars, you have all of the unified architectures built into it, whether that's unified architectures of the fabrics, it's the operations from a Nexus Dashboard perspective, and then bringing in intelligence of the services, and that you'll be talking about more, which is around the Hypershield and the services that we bring to bear on that, on these devices. How do we take all of that and embed that into a data lake, Splunk-based digital resilient data lake. That is our approach. Now, one thing that from a silicon and systems perspective, what we've done is we've done a traditionally, we've had networks with NPUs and forwarding planes on those switches.
You had those switches essentially build out with the architecture that takes you to a centralized service domain where you have the equivalent of what we call a data processing unit on firewalls, centralized firewalls. You build big honking firewalls, whether you need the capabilities or not. IP, layer four services, you may not need more than that. Now what we're doing is we're bringing all of these together under one common real estate with what we're calling the smart switches. You have the NPUs, which is Cisco Silicon One, and you have the DPUs, which is with our partner AMD, who have a booth right there as well. By the way, the smart switches are also here. You can see them on Plexiglas, right at both at the Cisco booth as well as the AMD booth.
What you're able to do with this is separate lifecycle management, not changing your operational paradigm, but common real estate, right? That allows us to deliver that infusion of security into the network. By doing this, now you're able to add services at an asynchronous speed, asynchronous capacity, and capability as well. That is how the smart switches come into the picture. We actually announced the Cisco smart switches, Nexus smart switches at Cisco Live EMEA. We've started shipping them. We have two form factors. One of them is shipping today, which is the 24-port 100 Gbps that you see there, as well as we have a 48-port top-of-rack form factor as well, 48-port 25 Gbps with six ports of 400 Gbps uplinks.
The use cases that we are able to support are these co-location use cases where you can on-ramp onto cloud or to other kinds of data center environments, zone-based segmentation where you do not necessarily need all of the bells and whistles of a full firewall. What you get much more is a stateful hardware-accelerated segmentation capabilities. What we did with ACI and the Nexus fabrics before, we are now moving from stateless to stateful. Fundamentally, we also add the DCI data center interconnect capabilities. That allows us to simplify your architecture both from an architecture perspective, but also from a business perspective as well, where you move from six devices, a few routers, a few switches, and some firewalls as well, and consolidating all of that into a couple of switches. That is a fundamental thing for those use cases.
Craig, this kind of architecture requires a fundamental rethinking of our designs and everything. How do we go about this?
Yeah. It's really cool to do this, to collapse four switches and two firewalls into two boxes. That's not enough. That's not really reimagining things. That's just packaging some hardware together. What we really need to do, if I don't click one extra slide, is reimagine the way that this fits into our policy models as well. The first service that we're launching on top of these smart switches is Hypershield. This is what our customers are testing today. The way that Hypershield imagines rethinking fusing security in the fabric of the network is by building a visual and data representation of your applications end-to-end using what we call our hierarchical data graph.
What this means is all the way from an agent that's running on a workload that's using eBPF to ingest telemetry to understand the originating function call, system call, CLI command, whatever happened to trigger an application behavior, all the way hop by hop through the network, ingesting network telemetry from these smart switches. We give you this end-to-end view of what triggered an application's behavior to happen, what are all the intermediate devices that it traversed along the way, and what are the policies that you may need to protect this application at each of those hops. We are not just putting the firewall inside the switch. We are also automatically learning where it is topologically. We are automatically recommending what policies you need to enable layer four zone segmentation using the Hypershield inside the switch. We are showing you how that fits inside your entire application domain.
This becomes a really rich source of data that we haven't had before for visualizing how applications are traversing our networks end-to-end and how they're behaving. When we think about tying that with Splunk and what we're doing from a data analysis perspective in the SIM and the SOAR, this data set becomes another data feed. Splunk, you may have heard about federated analytics. Federated analytics allows Splunk to pull data not just from the Splunk cloud itself, but also access various different data sources that may be running on-prem or in different places. Those could be third-party data sources like S3, like a security lake, like your own on-prem data lake. Those could actually be Cisco data sources. The data from Nexus, the unified Nexus Dashboard, data from Catalyst Manager, data from Hypershield, like the graph.
Now imagine that in Splunk, that graph that I just showed you that's showing how applications behave hop by hop through the network is paired with all the other threat intel telemetry that exists in Splunk. When an attack happens, I can see exactly every application that was implicated in the attack, how their behavior changed, what was the reach of the attackers into my network, and what may be compromised all through this unified data view in Splunk that hasn't been possible before. This all fits as part of our hybrid mesh firewall story. Hybrid mesh firewall, you hear that term a lot from analysts where they're just talking about unifying different types of firewall management together. For us, hybrid mesh firewall is really a rethinking of how we do firewalling fundamentally.
That means bringing together Cisco Secure Firewalls, bringing together within Security Cloud Control the ability to manage third-party firewalls, Palo Alto, Check Point, Fortinet, if you want to do that, all from a single pane of glass. We bring in our traditional micro-segmentation solution with Secure Workload, and then Hypershield, the two types of Hypershield agents that I talked about, the eBPF agents powered by Isovalent runtime security running in the workloads, the network enforcers running inside the smart switches that are working together as part of that Hypershield fabric. Now here, this week, we're adding to this ACI as part of our unified Nexus Dashboard and the Cisco Secure Router, which can also get security policies pushed directly from Security Cloud Control.
All of these different enforcement points, heterogeneous enforcement points distributed across the network, managed from a single policy in Security Cloud Control that can push down policies to all these enforcement points, and all connecting up to Splunk or whatever OCSF-compliant SIM you should choose to unify all of the data in a single place. This really allows us the biggest breadth and depth of blending security into the network. These are all really thinking about networking from a traditional speeds and feeds kind of sense. Networking today, especially in the age of AI, goes beyond boxes into the world of Kubernetes. The other thing that we have as part of this is Cilium from Isovalent. Cilium is the de facto standard for Kubernetes networking and security. It is the default CNI in every major hyperscaler.
What we have done is in that unified Nexus Dashboard where you have Nexus and ACI together, you can also see visibility into your Kubernetes networking with Cilium. This becomes, again, part of that networking fabric. That means on-prem or cloud, VM or Kubernetes, you have one consistent way of managing security policies, one consistent way of consuming telemetry and using that telemetry, and one consistent way of viewing your entire network fleet regardless of where your applications are running.
That's amazing. That's amazing. Also in that particular case, you're talking also about potentially being able to do the Live Protect capability as well with that?
That's right. One of the things that I talked about in this slide was, if I can go back, Isovalent Runtime Security with Tetragon.
Isovalent Runtime Security is an eBPF-based library that allows us to extend the functionality of the kernel at runtime safely for security reasons. In Hypershield, we use this for a couple of things: for writing segmentation policies inside your workloads themselves, and also for what we call distributed exploit protection, where we prevent vulnerabilities from being exploited. The next Log4j that happens, we can actually dynamically prevent in the kernel that vulnerability from being exploited. You heard today from Jeetu in his keynote about Cisco Live Protect, which is coming out first on the Nexus switches. Live Protect is also based on Isovalent Runtime Security. The same technology that Hypershield uses to protect against vulnerabilities in customer workloads, we're using to protect vulnerabilities inside Cisco's own appliances. This will extend to Catalyst switches and to Secure Routers and other devices.
The vision is you will never have another vulnerability in a Cisco device that requires an emergency patch. We'll be able to protect it from exploitation, and then you can patch during your normal maintenance window without your hair being on fire.
That actually is very, very pertinent to the data center, Nexus portfolio, for example, ACI fabrics and so on. Actually, any data center environment where you're dealing with the pressure of having to upgrade because of either a CVE or a Log4j kind of an issue, if we could protect that dynamically without having to create maintenance windows and all the pressure that's associated with that, that's amazing. This is incredible.
Yeah. eBPF, the underlying technology behind this, eBPF was co-created by Isovalent and Meta. They actually had different aims in the beginning.
Isovalent was looking at how can we optimize Kubernetes networking? How can we build a CNI without a sidecar? Meta was actually looking at this problem specifically because when a vulnerability, a critical vulnerability affects Meta servers, now think about how many servers are powering Facebook and Instagram and WhatsApp and Metaverse and everything else that's running in their cloud. It takes them 14, one-four weeks to roll out a patch across their entire fleet of servers because they do rolling upgrade cycles where they never take down too much of the capacity at once. They actually co-created eBPF so that they could do this on Meta's own servers and allow them to patch vulnerabilities at runtime without disruption. We brought that first with Hypershield, the customer workloads. That's awesome. Now we've brought it to Cisco devices with Live Protect working with the Nexus team.
That's awesome. Yep.
I think with that, we can invite our friend Neil up here and have some Q&A. Please welcome Neil Anderson, VP and CTO of Cloud, Infrastructure and AI at WWT.
Hi, guys. Good to see you. Hey, Craig. Great to see you.
I think what we're going to do is we're going to exchange a few questions both from our side as well as from Neil's side about what this means in terms of the fundamental disruption of enterprise design. Neil, if I could just start, what are the big primary challenges that your customers face when transitioning to this kind of an environment from a data center perspective?
Yeah. I mean, I think the big three challenges that our clients have is really around complexity, security, and then AI disruption. And we see that all over the place.
How do I secure my data center now, especially at AI speeds we're talking about, but still make that simple to do? It's just a huge challenge. I'm really excited about what you guys are building because it's just going to make it that much simpler for our customers.
Absolutely. How is WWT supporting customers as they make this transition? Obviously, you've trialed this in your environment. You have labs. Maybe you could tell us a bit about that and also maybe highlight an example of a successful data center evolution or modernization strategy that you've seen from a security point of view.
Yeah. We're very fortunate that we have invested in a lot of labs, and we've been working with Cisco on Hypershield and this architecture for quite a while together.
Our clients can come in, they can take a look at their environment, we can mock that up with them, they can understand not just the technology selection and how they're going to roll it out, but also what are the migration steps going to look like. That path to success for our customers is being able to map that out. We also have a special enclave part of that lab we call our AI proving ground. That is exactly what it sounds like. It is a specialized AI, high-performance networking, high-performance compute with Cisco and NVIDIA. That has been some of the announcements on stage by Jeetu today. We have been working together on that architecture, high-performance storage as well. We can bring that together for customers and then again, help them understand what does that migration path look like for me.
Thank you. Yeah. Yeah.
I mean, obviously, data center transformation is a big challenge. AI is driving workloads. I'm obviously interested in the security piece of this as well. How do you guys think about bringing security into this retransformation of the data center?
Yeah. We have a special, again, kind of enclave within that lab that's focused on AI security. How do you do that at the scale and the speeds we're talking about? It is really, really challenging. One of the reasons I'm really kind of excited about Hypershield is when you look at the current architecture, the way we've been building security for I don't know how many years now, 25 years, it's been this idea of a choke point firewall. I build a perimeter, I funnel all my traffic back to it, and hopefully I've got the capacity in that firewall.
We can really not continue that way. If you think about all the data that's going back and forth between models, clients, agent-to-agent communication, you've really got to push that enforcement much farther out towards the edges of the network. That's why we're really excited about Hypershield.
Yeah. I always say we want to change security from being a place in the network you send traffic to being part of the network itself. I'm glad to hear you say the same thing as me in different words. You mentioned Hypershield. You mentioned deploying AI. Now I'm just going off script because we have extra time and you're a really smart guy. AI for security and security for AI are almost at odds. We want to roll out more AI and we need to secure it. We want to use AI to secure it.
How are your customers thinking about those sort of competing priorities?
Yeah. That's exactly the way we talk with our clients about it. It's using AI tools for securing things as well as securing AI itself. That's really important. Cisco AI defense is a huge piece of that for us. Being able to put guardrails around models and the open source things that are out there, it's huge. That's why you're partnering with NVIDIA on that. They're embracing that to be able to secure their models and their AI stacks. It's hugely important. Being able to leverage AI tools for Sox and others to be able to get their arms around what's going on in real time. I was really surprised and pleasantly surprised at the AI Canvas that was unveiled today on stage. That's going to be just a phenomenal tool.
Think about a network operations team now can get that real-time population of what's going on. The same thing, I envision that you're going to extend that to SOCs and be able to have security engineers look at that from their lens and say, "All right, what's going on? Where are these threats coming from? What's the best way that you recommend to mitigate that threat?"
I think it's really fascinating the way we're reimagining what we do with AI, like AI Canvas, generating UI on the fly. I was reading some reviews of WWDC announcements yesterday, and Apple released this new liquid display kind of thing. It got panned a little bit on Twitter because it was very difficult to view in light mode on a screen.
I saw an interview with an Apple designer this morning who said, "Why are you thinking about looking at a screen? The future is not a cell phone. The future is much different types of visual medium." Right? I think it applies here as well. It applies here as well. We're rethinking the way that we do security. We're rethinking the way that we show UIs. We're really rethinking everything. Are your customers ready for this big transition that we're going through?
I think it's going to take time. It's a different kind of experience and a different framework for how do you troubleshoot, discover, how do you secure. It's going to take some time for people to rethink that. Right? I think once you're into it, I think it's going to become evident why it's so productive.
There is a great quote that I love to borrow from the CISO of FedEx. He says that AI is not going to take people's jobs. People that know how to use AI are going to take people's jobs. People are going to have to get really proficient at it on our clients, or there will be somebody else that does get that that's going to come along and be able to succeed.
Neil, one question that we did not touch upon is, as you see the customers, our joint customers, deploy infrastructure and as power and cooling become ever more important, how does an architecture like smart switches where you are bringing in security and infusing that as opposed to help in sustainability, which you obviously are building out for when you build the plan for your customers?
Yeah. It is definitely a challenge for customers.
These AI stacks consume a lot of power. They also produce a lot of heat on the other side of it. How do you power and cool that? Most of our clients, their data centers are not prepared for that. We really need to bring out every tool in the tool bag to be able to solve that, including products that can consume less power from Cisco, whether that's networking, whether that's compute and other things, firewalls, being able to get rid of that firewall appliance and actually embed it in the network. That's just a power savings. Another one. They add up. Right?
And real estate as well.
Exactly. You can get that power and cooling back so that now you can deploy AI in a reasonable manner. It's definitely a challenge.
If I could quickly plug something, we actually have liquid cooled switches being displayed there at the back. It's incredible. I mean, I saw it for the first time today where you actually have direct-to-chip liquid going, flowing in through the switch and coming out, and that consumes out the heat that's generated from the silicon and the various other components that are there. Similarly, we have the smart switch display as well.
Yeah. It's been a niche technology for a while. Right? It's nothing new. It was mainly consumed by the high-performance computing people. Now it's going mainstream. To actually cool these things, we're going to need liquid cooling for the kinds of GPU workloads that we're talking about here.
Absolutely. Yeah. Now, Craig, did you have any questions for us? Do you have any questions for us, Neil?
Yeah. One question I had, if I'm allowed to ask back to you guys, is how are you going to make this easier for our clients? You talked about segmentation, Craig. That's been really challenging for customers to turn on. App dependency mapping, it's been super tough to actually segment data centers. I'm curious, how are you working to simplify that? If I may, this also edge AI processing is going to happen. We're already seeing it. When you think about the single shot to a model, I enter a prompt and I get back a response, kind of nice and tidy client-server flow. We're moving into a very different phase of AI with agentic and physical AI that's really going to require edge processing to take place. It's not going to be tidy. It's going to be agent-to-agent, agents talking to agents.
You're going to have to secure that as well. Question back to you guys is how are you going to make that easy for customers?
Yeah. I'll start with the segmentation piece. Interestingly, it was like a year ago that we announced Hypershield and we said, "We're going to put AI agents in your network and they're going to automatically control segmentation rules." People were like, "AI agents in the network? That's crazy." Now it's like the only thing. Chuck said 50 times today on stage, "AI agent, Agentic AI." Hypershield, we set out to solve sort of the core problems of segmentation. The first is that how have we always done microsegmentation? We analyze applications over a period of time. We learn what good looks like. We put a deny-all rule at the bottom.
We put allow-rules above, and then we wait for the application to break because behavior changed or a new software update came out. Right? That is why segmentation that was already struggling in the past, when we get into things like Kubernetes, now we have AI-assisted development allowing us to move faster in writing code. Things are changing more and more rapidly. That model just gets harder and harder to sustain. The second piece is we're doing this with network rules, with stateful firewalls. Guess what attackers learned? We're using stateful firewalls. 80% of lateral movement we see is hiding inside of known ports and protocols because they know that we're going to open RDP and we're going to open our SSH and we're going to open MySQL and we're going to open all of these ports in our L4 firewall.
What we did with Hypershield, I showed the graph. The idea behind the graph is learn those behaviors automatically, but not just the behaviors. We're feeding with the eBPF events actual contextual information about what's going on. Who's making the call? Are there patterns of malicious behavior? That deny-all rule at the bottom, instead of just blocking anomalous behavior, it's actually reasoning about whether it should allow or deny that behavior based on everything it understands about the traffic, the application, the usage patterns. Think about a common use case I used to like to say to people, if you deploy Hypershield in dev, pre-prod, and prod, the Hypershield graph running in dev sees a new version of software show up through your CI/CD pipeline and a new port is required. You're very liberal in allowing ports to be open dynamically because it's dev.
Then in pre-prod, it sees the same version showed up and required the same port. That makes sense. Now in prod, it sees the same version showed up and it required the same port. It can tell the difference between spontaneous anomalous behavior in production and something that just worked its way through a CI/CD pipeline from a less restrictive environment to a more restrictive environment. Really designing this for the way that we build applications today. We are doing this a lot different than Cisco has in the past because we have been co-developing this with customers, much to the frustration of some of you who want to deploy it in production right away. I will give you a great example of how we got something wrong at the beginning that we had to redo in Hypershield in the course of development.
I mentioned we built it on top of ISOVL and runtime security. ISOVL and runtime security was built to protect applications, to monitor for security events, to block vulnerabilities from being exploited. It was not originally designed to be a segmentation tool. What we learned was we deployed this in a simulated production environment of one of our large healthcare customers, and it took 8,000 rules to segment a simple application because it did not support FQDNs and it did not support chaining subnets together, and it did not support a lot of the things that we would just naturally do if we were writing microsegmentation policies. We actually released first visibility-only auto-segmentation, and we went back to the drawing board and we introduced something called the Tetragon network policy.
You can see right over here a demo of this at work where we took Cilium network constructs, put them in the Tetragon policy, and now two orders of magnitude fewer rules to do network segmentation. Now we have autonomous segmentation working, and instead of 8,000 rules, it takes 100 rules. It's a much more effective solution. Right? Yeah.
You're bringing AI to make it easier for customers, which I love. I was hoping that I would see segmentation solved in my lifetime. I feel like I've been working on this for 25 years. I'm really excited to see this come together and the co-development as well. You guys have been great. We've been testing this together in our labs. I love that kind of very different approach to product development. Thank you. Thank you so much.
We've got 30 seconds left.
You mentioned edge AI. I want to plug a couple of things. One, every new Cisco device is a smart device. They all have hardware inside for us to run use cases. Hypershield is just the first one. It could be AI agents tomorrow. It could be anything else that's running in this hardware to help secure your networks or autonomously configure BGP peering or whatever it may be. Also, with UCS, our expandable chassis that allow you to dynamically add more blades of compute, more GPUs, and grow as your AI use cases grow. Thank you very much, Neil, for the questions asked.
Absolutely.
Thank you, Neil. [audio distortion]
Welcome back to the Cisco TV broadcast studio. We are so happy to have all of you with us here on the live stream. I'm Steve Multer. We have just come out of another fantastic center stage session.
This time, we heard about how data center security overload is helping us to architect networks for AI scale. Think about what we just experienced here, right? How Cisco is transforming network architecture by integrating advanced security and services directly into the network fabric and why that is so important, why that is really a game changer. We heard from Craig Connors and we heard from Murali Gandluru. They talked about how this innovative approach is optimizing our performance, how it's strengthening our security, helping to deliver more actionable insights, how it's helping to reduce costs. They also brought in Neil Anderson from Worldwide Technology to give us that behind-the-scenes story on how Cisco Hybrid Mesh Firewall and Splunk Distributed Analytics are coming together to help Worldwide deliver a faster network, but also a network that's more secure, that's easier to manage. Really, really exciting content.
What we're going to do right now is head out into the world of solutions where Michelle Moreira has caught up with Nick Consolo. Michelle, I'm going to send it on out to you, my friend.
Oh, Steve, it's so good to hear your voice again. All right, I'm on the show floor with my friend Nick Consolo. He is a Fellow Product Marketing Manager. This time we're talking about Cisco. There have been some incredible innovations this week with networking and security. For example, the N9300 series smart switches, as well as Cisco Hypershield. They both make it possible to fuse security into the data center network fabric. All right, Nick, share what this means to customers and how does this work.
Yeah, absolutely. Thank you for having me.
As you know, the AI-ready data center, when you're looking to pursue AI, whatever it is in your organization, networking and security are critical, especially having high-performance networking and performance there. What this means for your organization with an N9300 smart switches and Hypershield is we're allowed to apply stateful layer four segmentation within those smart switches. What does this mean for your organization? Essentially, what we're being able to do is provide that high-performance throughput that you would have traditionally relied on for data center firewalls, where you were not requiring that layer seven or advanced threat capabilities, where you were typically relying on the performance instead. For those data center interconnect scenarios, there's zone segmentation or your cloud edge for cloud on-ramp.
We can now provide you 800 Gbps of throughput through a smart switch and provide with Hypershield embedded that stateful layer for firewalling. Let's go ahead and get into the demo and show you how it works. From here, we start in Security Cloud Control. This is our unified management platform where you can manage all of your SaaS-based Cisco security solutions, as well as your third-party firewalls and Cisco Secure Firewalls. If I were to go over to our network-based enforcer, we can see that we have our Nexus smart switch configured with our network-based enforcer with Hypershield on it. If we move over to the policies, we see that we do not have anything in there right now. We are working with a clean slate.
What we're going to do is open up that TCP and UDP connection going from our clients to our server and having it move through the switch by opening it up with a policy. From here, we'll start with a performance test that shows you that this traffic is not passing through the switch. It's being blocked with a default deny as we have that traffic routing from the clients to the switch. We'll open up and create that policy. We do this by using our park model based on AWS Cedar, where we can describe the intent and allow the system to build and perform the policy for us. In this case, I want to allow my client to communicate TCP and UDP to my server. We'll go ahead and deploy this.
Just hit refresh here and see that our policy has been deployed. If we were to go and look at that performance test that I was showing you earlier, we now see that almost nearly 800 Gbps of throughput is moving through our switch from our clients to our server. We have now established that stateful layer four segmentation connection. Now let's take it a step further and actually show you blocking UDP policy. Before, we'll show you a baseline test, this one showing that everything is operating as we showed earlier, but we're going to want to focus on the security transactions here below in just a moment. We go back and create this policy. As we create the policy, I want to block UDP communications that's coming from my client that's trying to reach my server.
The beauty of this is now we're going to verify that policy. What we're doing here is replicating this traffic behind the scenes and applying that Hypershield enforcement policy on that replicated traffic. We can test that traffic in intervals. In this case, we're doing a quick test, but we're recommended for 30 minutes. You could do it as long as you want in order to build confidence in that policy before you ever hit deploy. I'll show you what I mean. We'll go ahead and verify that policy, hit refresh, see that the policy has been verified, and we'll go down to the verification results. We can see that compared to our original, before we had our original policy, which was opening up that TCP and UDP.
This is the new draft policy that we are looking to verify, seeing that our hits drop just different. We had a small hit count, which means that the policy is doing as intended. We will go ahead and click deploy. Once we have hit deploy, we can refresh and see that this policy is now, as we looked at that baseline, our baseline is showing the UDP and TCP as I was building the policy. Once I enforce the policy, it is now dropping the UDP traffic, establishing that stateful layer four policy enforcement. We can also corroborate that with the security mix transaction showing that we are blocking that UDP traffic. What does this mean for your organization?
We were able to show you that we're pushing 800 Gbps of throughput into a Cisco N9300 smart switch, which takes up a two RU rack space. Usually, that would take an enterprise-level firewall to be able to manage that level of throughput. We've also been able to enforce stateful layer four segmentation across that switch as well. Being able to provide you those firewalling capabilities and allowing you to consolidate and rethink your segmentation strategy by consolidating your data center architecture and doing it with a significantly lower total cost of ownership.
Nick, that was incredible. This is the deep dive that we get here at Cisco Live. Please come down, see it for yourself. Now, Steve, back to you.
Thanks so much, Michelle. Give my best to Nick. I haven't seen him since RSA conference. Great overview there on Hypershield on C9000 series.
Great, exciting announcements here at the show. In yesterday's opening keynote, we heard Chuck say that the network is critical, right? The race is on. That is because our speed demon transformation right now toward Agentic AI has really made the network more important than it has ever been before. We have been talking about that all week. By Cisco making security a core native building block of the underlying network, it assures that we can apply Agentic AI applications across the network appropriately, intelligently, ethically. This is really core stuff, and I am glad that we are having these excellent conversations. Speaking of which, Lauren is here in studio with Joe Vaccaro talking network assurance and monitoring. Lauren, let's send it over to you.
Hello, hello. Yes, I am here live with Joe Vaccaro, Vice President and General Manager of Cisco ThousandEyes.
Joe, why is it critical for customers to really rethink their current approach to network monitoring?
Yeah, thanks for having me. As we think about these transitions that our customers have gone through, that rise of cloud computing that provides a new place where our workloads are running, the rise of an internet-centric network architecture that's really transformed the way that connectivity happens. Even if you think about the adoption of SaaS applications, and now we're entering this new chapter about the introduction of AI. What's common across all four of those is that our customers rely upon them to run our business, but they don't own them. If you don't own them, how do you assure the experience through them?
We know that these digital experiences really power our customers, whether it's the customer trying to get to your business, whether it's your employees trying to access critical applications. This is paramount to how businesses operate. If you don't own these critical services, you need to be able to see them and understand them and build and prove them like you own them. That's really at the core of what we do inside of Cisco and what's really powered ThousandEyes over the last years being part of Cisco and the overall family.
Amazing. I want to just tag a little bit on just assurance because I heard Steve mention assurance. We've been hearing a theme of assurance throughout the conference. What does that really mean? Why are we taking that so seriously here at Cisco?
Yeah, because if you think about what does assurance mean, by definition, assurance is a promise of an outcome. Applied to our customers, what that means is how do we help them at Cisco to be able to deliver on that promise of a great experience? What does it mean to deliver on that promise? First off, you need to be able to detect issues before they impact your business. You need to be able to localize them across this complex digital supply chain between the device, the networks, the cloud, and the application. You need to be able to diagnose and understand why did that issue happen in the first place? You can begin to resolve that problem.
When we think about how we're modernizing assurance and helping our customers move from this reactive approach to this proactive approach, assurance helps us stay focused on solving those most complex customer problems.
How is Cisco uniquely equipped to do just that, especially for owned, and you mentioned it earlier, unowned networks and services?
Yeah, because digital experience has no boundaries, right? They traverse from any device. What is core to our strategy and really what sets Cisco apart is three things. One, can you be able to measure from everywhere it matters to that customer? Can you see from any different type of device out there? Cisco sits on every different type of environment, whether it's a collaboration or a video conferencing room in your office, whether it's a switching and routing, whether it's the secure client running on any different type of device.
From a ThousandEyes perspective, we allow you to be able to gain that key visibility from all those different types of devices. That breadth extends also in being able to see through the entire path, deep understanding what's happening at the device, being able to go deep through the wireless network, the wired network, the wide area network, the internet, the service provider's network, the cloud infrastructure itself, and even into the application. Now, that alone is not enough. We do all of that today, but it's also the depth, being able to understand deeply what's happening across that problem. Again, you can be able to gain that level of intelligence to quickly resolve those issues.
Those three things, the ability to see and measure from everywhere, be able to light up that entire path, and to be able to bring context as a source of where you need to solve that problem is really central to what sets Cisco apart in this industry. [audi o distortion] to unpack new vantage points where you can be able to run your visibility from, including mobile agents, industrial IoT devices, being able to see deeper through the connectivity, including support for Cloud Insights on Azure that allows you to be able to see the Azure set of services as if you own it, as well as integration into Splunk that allows us to be able to see deeper through that application.
Then also the integration of the Cisco AI Assistant to help provide that augmentation that uses the power of AI to be able to help our customers through these elements. Then using ThousandEyes Intelligence deeply integrated into that AI Canvas, I'm really excited about what that's going to bring, that generative UI experience that'll help our customers to really see a holistic view and, again, focus on solving problems, delivering on that promise of that great experience.
I love everything you said, and just the key part being that it's all one Cisco. We are really bringing this together and making it easier for all of our customers. That was amazing.
Yeah, because that experience knows no boundaries, right?
From one Cisco, we can be able to empower our customers no matter where you're measuring from, what you're measuring through, or how you want to be able to solve that problem. Cisco's here to help our customers.
Thank you so much, Joe. It was an absolute pleasure. I am going to pass it over to Steve here.
Thank you so much, Lauren. Joe, great job. Again, it's about the power of the platform. We keep saying that Cisco is so uniquely situated to combine the power of the network with the power of security, with the power of the data from Splunk. Just really, really terrific, terrific insights. All right, we are headed back out to center stage in the Cisco showcase. This is going to be an exciting one.
Kelsey Doran, Cisco Strategy Director and the Chief Sustainability Office, is going to share some strategies to help us respond to rising costs and to regulatory shifts, to increasing pressures to meet sustainability and compliance requirements as AI and other factors continue to drive us and drive up energy demands. Kelsey is also going to welcome Steven Santini, Global Strategic Alliances at Schneider Electric, to show how businesses exactly like theirs, like yours, like all of ours, can meet that energy challenge and turn it into a competitive advantage, helping us leverage digital innovation and drive energy efficiency, more cost savings, help us drive growth in our emerging AI-powered territory. All right, again, glad to have you with us. Here we go.
Pretty cool. Let's go ahead and welcome our panelists to the stage now. Come on.
Hello, everyone. My name is Vijoy. I run this group called OutShift inside of Cisco. It's the internal incubation engine for Cisco. As part of that charter, I also have Cisco Research, Cisco DevNet, which is the Developer Relations Organization, and Open Source. The charter of OutShift is to build products that are adjacent to Cisco's core markets. Adjacencies could happen in terms of personas. We always talk to network engineering persona or SecOps or AppOps, Webex, and collaboration. Are there products that we can build that are adjacent to those spaces? As well as, can we build products which are a bit further out and are a little riskier when it comes to technology or market? Clearly, the topic of today, quantum, is in that space, which is these are technologies that are a little bit further out, but they're happening faster than one can imagine.
With that, I'd like to introduce the panelists here. Let's start with Anthony. Go ahead and introduce yourself.
My name is Anthony Grieco. I have the pleasure of leading the Security and Trust Organization. Our mission is to help protect Cisco and work with all the products that we produce to help make sure that they're ready for the future from a security perspective. In particular, we'll be talking a little bit about post-quantum crypto and how we protect information in the age of quantum computers.
Bill.
Hi, everybody. I'm Bill Gartner. I have responsibility for our Optical Systems business, which are the systems that are used to carry traffic over fiber optic networks that are traversing a city or a country or even subsea.
I have responsibility for our Optics business as well, which are the optics that are used inside a data center or central office.
Ramana. Yeah, my name is Ramana Kompella. I lead Cisco Research. I'm a Cisco Fellow as well. My team actually conducts research in emerging technologies of strategic interest to Cisco: AI, networking, cybersecurity, what have you, and the topic du jour, which is quantum. Really excited to be here.
We have everything from research to product to whether we can deploy it securely in the organization. We're going to have a fun conversation here. What I wanted to start with is, in general, show of hands, how many folks out here think that quantum computing will be a reality in the next five years? Wow, that's like a... Optimist. Optimistic crowd. I was like 70%.
How about five to 10 years? I think that's the remaining 30. And 10+? We need to convince the three of you. Let's spend some time. There is some real artifact just behind this stage, or in fact, in front of it, so we can walk over there and there are people who can explain what we are building. Let's get to that. I think I'm an optimist. A bunch of folks here are optimists, but we're also realists in figuring out what we can do today and what we need to solve for making that five to 10 year roadmap a reality. You hear a bunch of news all the time, just like AGI. I mean, I think this show has been all about agentic and AI. People talk about when will AGI or artificial general intelligence come to the scene?
The real question is, does it matter? Because yes, there is some definition of AGI. People can't even seem to agree on it. That'll happen sometime in the future. We have pragmatic use cases for AI, for Generative AI, for agents even today. I think of quantum in the same way, which is there might be a time in the future where pragmatic scale-out quantum computing might happen, but there are real-world use cases that are being discussed today that we are trialing today that we should talk about and we are all excited about. The other thing I want to just phrase upfront is, whereas everyone else talks about quantum computing and compute nodes, these are the servers that can actually do quantum computing and run those algorithms.
We are in the business of building the secure networks that connect all of these quantum compute nodes together to form what is known as distributed quantum computing. We are Quantum Networking and quantum Security space, and that's what we are primarily going to focus on in this conversation. With that, Ramana, I'm going to come to you. We just announced the Santa Monica Lab. It's at the nexus of a whole bunch of research that's happening in Southern California. There's Caltech, there's UCLA, there's UC Santa Barbara, and there are labs from Amazon, Microsoft, Google, and Cisco in Southern California. What is it that we're doing inside the Santa Monica Lab?
Cisco's quantum research program, the goal of it is basically to build a world-class Quantum Network of the future.
The Santa Monica Lab, the quantum lab, is basically where we have our small but mighty quantum team led by Reza, who's actually in the audience. They conduct research in developing the next generation of Quantum Network hardware, software, switches, entanglement sources, all kinds of stuff. You can see some prototypes out there in the OutShift booth, which are a result of all the hard work that the team has put in over the past several years. This is just the start. I mean, there's a tremendous amount of Quantum Networking and to touch upon what Vijoy was mentioning, to truly transform quantum computing and pull in the timeline of practical quantum computing significantly. Quantum Network is a big part of it. This lab is all about basically doing the experimentation, research prototypes, all that stuff.
You know that, I mean, we talk about quantum computing and the practicality of it. If you know that quantum computers actually work on quantum bits or qubits, which are different than regular classical bits, can you give a quick 30-second explanation of what qubits are and how many do we need and where are we today?
Quantum Networking is fundamentally different from classical networking. In the classical world, you would think of, let's say, Alice wants to talk to Bob. Literally, they'll make a packet out of the zeros and ones, and this packet will journey through the switches and eventually reach Quantum Networking, on the other hand, is completely different. Quantum information transfer happens via a couple of key principles. One is called entanglement. The other one is teleportation.
Quantum entanglement is all about generating a pair of photons that basically share entangled states. Think of them as these Amazon boxes like Alice and Bob have, OK? If you want to transfer quantum information from Alice to Bob, literally, you put that information in the box that Alice has, and it gets instantaneously teleported to Bob. Now, it may appear like it's actually like breaking a bunch of laws of physics here, but it turns out teleportation happens where the information gets transferred, but in order to recover that information, you still need some classical information to go via the classical channel. As a result, no laws of physics are violated here, but that's the way it works. What we are really excited about is the entanglement source, which is really what we have actually in the booth right there.
The entanglement source that we have is a small chip that basically generates entangled photons at a very high rate. We have about 200 million entangled photon pairs per second. The other beautiful feature about this is when you think of quantum, right, you would imagine these gigantic cryo chambers and super low temperatures, like close to absolute zero. This chip actually works at room temperature, and it's compatible with existing fiber technology, which makes it super, super practical and easy to deploy. The rate at which we generate these photons is tremendously high, and we will actually increase it substantially further as well. This chip actually is a foundational technology. We need several other components in order to truly enable this Quantum Network of the future, which are basically prototypes that we're building in our lab.
Stay tuned for more updates in the coming quarters and months.
One of the problems that we are going after first in the quantum domain, at least the vision, is that we build, again, these distributed quantum computing environments so that you can connect smaller quantum nodes, which are right now 200 qubits, they're really tiny. For practical computing, you need maybe hundreds of thousands or even a 1 million qubits. By connecting a bunch of these smaller nodes through this fabric, you can accelerate that timeline for pragmatic quantum computing. The data center is a restricted enough environment that you can deploy this and experiment with it. With that, Bill, coming to you, this is a research prototype, but to make it a reality, what are the challenges that you see from the product perspective to make it real?
First of all, let me say I think that this is really critical technology, and this team has done a terrific job of advancing the state of this technology. I think Cisco's birthright really is to own networking, and Quantum Networking. the idea of starting inside the data center by networking computing elements will give us a lot of insight about what it's going to take to build a larger scale network. I think this is a really important thing for Cisco, and it's a really important thing for the industry. We're at the research stage here, and many of us have lived in a world where you start in research and you move products into production, early, small scale, and then large scale. We're early in the life of that, but I think things are moving very quickly.
The entanglement source that you see here is a very important first step, and it's generating entangled photons. We need to take this into a production environment, so that means we have to certify a fab that can actually build this thing. As Vijoy and Ramana mentioned, we would like this to run at room temperature. If you've seen any quantum computers today, like IBM's quantum computer or Google's, those do not run at room temperature. They require massive cryogenic sources to keep everything cool, very cool. We would really like to have technologies that run at room temperature. Another key technology for us inside the data center is the switch itself. The switch has to preserve the quantum state. Today's photonic switches will likely destroy the quantum state. We have some technology development ahead of us.
I think the next step is to take this into production environments and climb yield curves, which we are very good at. We'll figure out how to make this a reliable solution that can be deployed in massive scale for our customers. That's all focused on what's happening inside the data center, which is a good test bed for us to figure out what it's going to take to build larger scale networks.
This is all great. You're telling me that it's actually closer than what the research guys are thinking about? I think. No, just go ahead.
Yes, I think it is closer. I'm more of an optimist. That's amazing.
Let's challenge you a little further.
Inside the data center is great, but what we are truly after as a company is to build out the quantum internet. There are two parts to that statement. One is, can we connect quantum data centers together to enable long-distance computing? What is also possible today is you have quantum sensors that are being used for weather prediction, better weather prediction, or for biomedical imaging, and so on and so forth. You need to build out a quantum internet which carries these quantum signals further and further apart. What are the challenges to get to the quantum internet, Bill?
Right, we have many challenges that have to be overcome in order to build what we would consider a WAN network. We will get there. As an example, distance is one significant issue for sending entangled photons.
Today, there's very limited distance that we can send an entangled photon pair and have it survive. We need to find a way to extend that distance with things like repeaters, repeaters that have to preserve the quantum state, and things like memory, quantum memories. Those technologies are under development right now. There are companies that we are working with. There's a very large partner ecosystem that's developing around this technology. Those technologies are under development, and we will see some of them coming out as early as this year. One issue is increasing the distance. That's a big issue. Another is increasing capacity. I do want to put that in context. The first DWDM systems that are used in classical networks were 40 gig systems, 40 gig on the fiber.
Cisco Live, the world of solutions will close in 15 minutes.
15 minutes to closing.
Forty gig on a fiber were the initial DWDM systems. Ten years later, we had increased that by an order of magnitude, 400 gig. In the next 15 years, we've increased it by over two orders of magnitude. We can now demonstrate 64 TB of capacity on a fiber. In addition to that, we've really done a great job of shrinking the technology. This is an 800 gig pluggable optic that can transmit a signal 1,000 km. That's not where we're going to start with quantum. Quantum is not going to be at parity with classical networks. As Vijoy said, there are going to be applications like sensing that can take advantage of short-distance applications, low bit rate applications, and deliver very significant value.
I think we're going to see applications emerge that sort of match the state of the technology over time. I fully expect, just like we did in classical networks, we're going to see massive improvements of the technology over time as well.
Thank you. I think my goal is to get to a point where we can carry that entanglement source in a pocket, just like Bill carries around that transceiver, because that is a metric to achieve, which tells you that we are at scale. Otherwise, we aren't. Coming to Anthony, I mean, we are all optimists here. I'm not saying you're not an optimist, but I think we are out there to build distributed quantum computing. One of the things that it'll do is it'll enable practical applications and algorithms to run on quantum compute nodes faster or earlier, rather.
One such algorithm that we are all paranoid about is Shor's algorithm, which ensures, or not ensures, but which actually enables you to compromise existing key exchanges that happen between computers. This is the whole notion of it can be done either today, or you can go after these store now and decrypt later kind of attacks, where you're snooping into traffic, storing them now. When you have access to these computers, the quantum computers, you can actually make all of that conversation plain text. By building these distributed quantum compute nodes, we are accelerating even that timeline.
Yeah, gee, thanks.
Yes, help me, please, because we are building the technology, but you need to secure it. What do we do?
Yeah, your optimism is the reality of security folks, which is it creates a set of challenges that we've got to start to think about now. You did a great job of describing it. I mean, the key risk that we think about that we need to begin to deal with today is the idea that you can capture traffic and when a quantum computer is created of sufficient scale, which we are looking to enable, that traffic is potentially then decryptable by an adversary. When you think about that conversation, many people think about that as a big government conversation. I think that's a really relevant set of actors in this space. You think about the importance of AI and how important data is to AI and will be in the future.
There's actually new reasons to want to do this for folks that may not be able to get their hands on all the data that they may want. They may want to choose a way to get after it. When we think about that classical problem, we've really been engaged today in standards work to help build quantum resistant cryptographic algorithms. That body of work is something that Cisco has directly contributed to. We've worked with NIST and other organizations to create those algorithms and make sure that they're out in the public domain. We're working constantly with standards organizations, in particular the IETF, to get those algorithms leveraged in protocols that we use today. Ultimately, thinking about the risk that we all need to be dealing with today, it is this encryption of data in transit and doing that encryption with quantum resistant algorithms.
Cisco has a robust roadmap across our portfolio that we continue to enhance and refine, which will allow us to implement quantum resistant cryptographic algorithms. In fact, the campus and branch launch that we've done, that hardware is enabled for quantum resistant algorithms. This is really essential for us to start to think about now. When I poll my peers and other security leaders in other organizations, to be really candid with you, most of them don't even have this on their radar as things that they're worried about. I tend to joke maybe they all think that they're going to retire before they actually have to deal with this. When we think about ourselves as stewards of where we need to be going and what we need to be doing as networking professionals, as security professionals, facing this reality today is super important. It doesn't end there.
It's not just about the transport algorithms and how we send information across networks. When quantum computers are created, all of the things that we think about, you think about signing an image. You know, how do you know that it is a legitimate Cisco image that uses public key cryptography? That in and of itself will be in jeopardy of being compromised. Somebody could create something that looks like a Cisco image, but isn't. There's risk up and down the stack of everywhere that you worry and use cryptography to protect security. We've got to be embracing the future state of where we need to be, quantum resistant algorithms in this space. We're really leaning into it heavily and have invested a lot. We also have a lot of ways that we can enable quantum key distribution inside of our products with standard protocols like SCIP.
This is a rich area that is going to drive what needs to be a fundamental priority for so many of our organizations. To be honest with you, most of us are just now getting up to speed of this is what we need to be going to deploy. This is a do it today, especially with this audience of 70, 30, and those time frames. You've got to be taking these things really seriously today.
When I think about the timelines and the phased approach here, I mean, there is education, there is visibility as to which services are probably most critical. You talked about encryption for software at rest or data at rest, but also in transit. How would you draw up that timeline? What should people do to prepare?
I mean, you've got to take a risk-based approach.
You can't fix it all at once. Focusing and getting an assessment done of what your most vulnerable communications are and taking that assessment and using that to inform your roadmap is a really essential set of things. Cisco has a set of services that allow for customers to go through that assessment today with our CX organization to understand where is the most risk and how is it that you would address it. It's most likely going to be in your transport layer, and it's most likely going to be in environments where you're traversing someone else's infrastructure. We have assessments that can help our customers do that. In fact, we've done that inside of our own enterprise, which is my job of making sure that we're prepared for this.
Anthony talked about security and what we can do with standards like the ones that NIST came out with, the post-quantum cryptography standards, which are all software-based. Ramana, I know that there are other use cases that we have for the Quantum Network itself. Even though we are building out the Quantum Network to enable distributed quantum computing, there are a whole set of use cases that are possible today in the classical world. Security is one of them. Can you just talk about a few of the others?
Oh, yeah. In our journey of actually building a Quantum Network, right, we started off with this, let's build a Quantum Network for enabling practical quantum computing, scalable quantum computing.
In our journey, what we found is that there are Quantum Networking use cases that Quantum Networking that can actually solve classical applications. That's actually really exciting because we don't have to wait for, I don't know, depending on your choice, it could be five years or 10 years. You don't have to wait for quantum computing to be real in order to realize the value of a Quantum Network. Some of these use cases are decision coordination. Decision coordination is around how do you have Alice and Bob take a coordinated decision? It could be stock trading. It could be something else. If Alice makes a decision to buy a piece of stock, Bob should also be doing it at the same time. How do you coordinate that?
The classical world, you'll actually have to send a message saying, hey, I'm actually taking a decision to sell the trade or sell the stock, and then that information would go to Bob. If you have a Quantum Network, it turns out you can leverage the entanglement property that I was mentioning before to allow for Alice and Bob to take the same decision at the same time. That's a fantastically interesting application. Similar other applications, we have secure position verification, ultra precise time synchronization. These are all use cases that can actually take advantage of a practical entanglement-based Quantum Network. Now, the fourth one that is listed here is eavesdropper-proof security. I'm sure you must have heard of the word quantum key distribution, which is leveraging quantum properties to exchange keys between Alice and Bob.
That requires a separate Quantum Network or a dedicated fiber to exchange your quantum keys. If you have eavesdropper-proof security built out of a Quantum Network, essentially that means that you can leverage the no-cloning theorem property of a quantum channel to detect an eavesdropper. You do not have to exchange keys. You just detect that there is an eavesdropper on the wire, and that will allow you to make sure that either you exchange your information or not, depending on what you want to do. These are the kinds of use cases that we can build with an entanglement-based Quantum Network today. We do not have to wait for the whole quantum computing revolution to really take place. That is really super exciting for us.
What you're saying is that even for the optimists in the group, which are saying five years out, you're becoming an even more exponential optimist, saying, let's start using the Quantum Network for classical use cases today because it's here and now. I already got your answer, so I'm going to move on to Bill. The question I had was, what is the one quantum opportunity that excites you today?
Certainly, security is a very clear opportunity. I think sensing is one application that is just magnificent in terms of the scope of what can be done with Quantum Networks for very, very, very precise sensing and solving problems that we just can't solve with classical computers.
Anthony, how about you?
For me, the idea that we could not just protect information as it's being transmitted, but detect when someone is listening or potentially trying to listen, that eavesdropper-proof piece of this for me is if you look, there's a lot of people that spend a lot of time trying to figure out whether or not their communications are potentially compromised. This, for me, is, I think, a real game changer in a lot of ways. That classical computing application of this is something that, before we started having these discussions, I didn't even realize was a real possibility.
I think that is the most paranoia-inducing use case that we can all line up behind.
The other thing, I was talking to somebody, and the decision coordination part of it, it's almost like, let's say there was a starting gun that could be heard by anybody across planet Earth at the same time. What could you do with it? The use case that was brought up was like, hey, I want to play a game with my friend in Tokyo. I don't want him to get an advantage on the gaming exercise before I start playing the game. He could be starting out first and get an advantage just because he's three milliseconds or a few seconds ahead of me. Can we have a starting gun which says, start your game now? I said, that's an amazing use case. Boy, you don't want to build out a Quantum Network just to start gaming. Maybe we do.
I'd be surprised.
Maybe at some point, when Bill scales it out, it's cheap enough that we can do that. With that, I would like to wrap up and say, as we heard today, I mean, yes, the vision is to Quantum Networking and quantum security from the ground up. It is not just that quantum entanglement source that you can just see behind you, and you can go and visit that booth. It is also the protocols, the software controllers, the compilers. In the end, you should be able to drop in a few lines in your Python code, and stuff should work. That is what we want to get towards. As we do that, we have use cases in the classical world today that we are super excited about.
Even above all of that is the paranoia and the realistic check that Anthony brings to the table. Start preparing for the, it's called a Q day, I think, or a P day. I think there's a term for this. I should know this, where quantum computers would be strong enough that they can break key distribution today. Start preparing for that because that's a long, long, long time.
It's called my retirement day.
That's your retirement day. Hoping all of us are retired by then. Thank you for coming. If you have questions, we'll be around. Thank you.
Welcome back to the Cisco TV broadcast studio. We just got a fantastic final word as we head toward the end of a powerhouse Wednesday here at Cisco Live 2025 in San Diego. What an amazing day it's been.
We just got some key strategies on how we can respond to rising costs, to regulatory shifts, increasing sustainability and compliance requirements, and the ways that we can leverage digital innovation to drive more energy efficiency, better cost savings, more growth in our emerging AI-powered world. We've got two more stories for you this afternoon. First, we're going to take one last visit down to the Cisco Showcase, where Michelle is bringing us the inside Quantum Networking with Reza. Michelle, let's send it over to you, my friend.
Steve, we may have saved the best for last. We're talking Quantum Networking with Reza, Head of Cisco Quantum Research. Reza, you just heard your colleagues talk about quantum timelines. Was there anything in that discussion that stood out to you, or was there anything that you wanted to get up there and correct?
Thank you for having me. No, all they said was correct. What I wanted to really jump in to say is that, I mean, if you look at the monitor, what's happening in the quantum technology domain, the big quantum computing vendor, a couple of them in the last two days, they updated their forecast to make a quantum computer reality for three to five years earlier than what originally was anticipated. It shows that what we do here to build the network infrastructure to connect them together, that's really, really important. We need to make this faster even than what we anticipated.
The other thing that I wanted to also add to that is that it's very interesting for me when we started this project at Cisco to see how beautiful is that what you can do, really bringing a group of scientists and a group of engineers together in the same environment to create a really to make something that maybe you think it as a science fiction, a reality that is really to go to a customer and be deployed in the field.
Speaking of science fiction versus reality, I have to ask about your lab in Santa Monica. Can you describe what this is? I'm thinking something from the movies. Does it have big machines, or is it just a normal engineering lab?
It's a normal R&D engineering lab. However, that's for the device. You see the technology is there.
The use cases that we demonstrate or applications that we demonstrate are coming really from science fiction movies. We demonstrate in our lab information teleportation. In a Star Trek, they do teleportation, but we do not do the human version of it. We do the information version of it.
OK, that leads me to what folks back home are probably wondering. The bottom line question: Is this ready? When can customers expect Quantum Networking?
This is in terms of a technology that is mature enough, but it is not a product yet. It still is an R&D prototype. We are working with several design partners that are working with Cisco to see how we can use this technology in their environment and for their use cases and what problem or pain point they have that we can address with this.
That comes from the quantum computing vendors to a more near-time application. For example, we are working with financial customers to see how they can use this network technology to even improve what they do in financial transactions or trading. We work with a community like robotics or scientific community that needs really high precision synchronization. We will use the same technology. We also work with the people that need really to have an extra layer of security using quantum technology. We can use this technology. It is not only about the quantum computer or when the quantum computer becomes available, which is very real, but also we can use it now for a specific use case or a specific pain point of a customer of Cisco.
No, that makes sense.
What about the past few days has really stood out to you here at Cisco Live in San Diego?
The level of innovation in Cisco is amazing. You can go to a sit an AI session and see, wow, what we are doing in AI. I am not an expert in AI, but I go sit there. I learn a lot. I see amazing futuristic things happening. You come to Quantum, which is my favorite, my baby. You see that we have created such a small chip. That small chip creates a 200 million pair of entangled photons that you can do that science fiction with teleport information with that. That's amazing. I mean, the whole stack of network from application to physical layer, we see innovation in Cisco Live.
Thank you so much. I had so much fun in this interview.
We did leave the best for last. Steve, I'll hand it back to you one last time today.
Bravo. Thank you so much, Michelle. And to Reza, I really do love the work that he and his team do at OutShift and at Quantum Labs. Really cool stuff. All right, for our final exclusive story to wrap the broadcast day, Robb Boyd had an opportunity to sit down with Martin Lund, Executive VP of Cisco's Common Hardware Group. They talked about new announcements around silicon application and control across Cisco's AI infrastructure, switching, routing, compute, Wi-Fi. Let's join Robb and Martin in that conversation.
All right, I'm so excited. We've got Martin Lund, now EVP of the Common Hardware Group, because we've talked—did I get it wrong? You got it right. You're giving me the look. No, this is fun. We have spoken before, and I always enjoy this.
Let me throw this out at you for just a moment. Silicon One, I feel like it's not a secret. It's been out for a while. There's been iterations. There's been lots of product integration. I feel like it's having a moment in the sunlight, so to speak, certainly at this conference. Over the last year, I feel like it's been building. What is your impression? First of all, am I accurate at all in that?
I think you're right. Silicon One was announced in 2019. Since then, a lot of work has gone into building many different types of variations of the silicon for different locations in the network. What we have done here, we're rolling out with new announcements also in the campus that is used in the campus. It's used for service providers.
Obviously, it's been used for the hyperscalers for a while. We're kind of pretty excited about coming out and really talking about it. If you look around at the show floor, there are lots and lots of places where it shows up.
It's not just in one place?
It's not just in one place.
I want to understand what the competitive advantage is here. First, I think you have one with you.
I brought one.
Would you share with the group?
Y es. Let's take a look at this we have here. This is a G200. It's a 51 TB of switching capacity in a single chip, 512 ports of 100 gigs on a single device. It has 64 ports of 800 gig or 128 ports of 400 gig.
This one cracks me up, is that you're talking about what I normally associate with a big racked switch?
That's correct.
It's in there. It's actually at the heart of probably the switch we've developed.
It is part of it. It not only does phenomenally great switching of packets, it also has a flexibility built into it because we use a programmable engine that allows us to adapt feature sets. Even after the chip is made, I can add new features to it. A ll that without paying for it in power. It's very exciting.
Cisco made an investment in this area, which is unique. Not very many people could even attempt to make an investment in their own silicon.
The way in which we use it and the way in which it forms a fundamental part of the go-to-market uniqueness that Cisco offers, how do you describe that? What's the importance of having this type of capability in-house at Cisco?
I think it's super important. It's super important for hyperscale customers. It's important for service providers and also for enterprises. As we're fusing security into the network, we need to control the core innovation elements here. We have built this architecture so we can add capabilities over time, like security, and be supportive of this unknown. What is the network going to look like with Agentic AIs? I can tell you one thing. It's going to look different. Yeah, and the demand and the importance of the network is going to just go up.
S o having control over the innovation tool kit so we can optimize it and also build in things that have not been built before is key. One more thing. These devices also get used into the data centers of hyperscalers. Hyperscalers have been an important thing for Cisco over the last year. Even more so in the last quarter, you have seen some updates that show that we are making good traction there. For me, that would be super exciting. I used to work at one. The other part of it is that it is also what allows us to compete in the marketplace in the most demanding workloads. That is the part I do like.
At least have some standalone value.
We offer—I do not want to go into this—I was just thinking of white box switches, the ability for us to provide an element of familiarity and confidence for customers who have different needs that may not involve our hardware all the way through.
Uniquely, because we have our own silicon, we also, when we announced it in 2019, we announced that we are willing to sell this as a chip, as a white box, or as a full-branded Cisco system with the Cisco software on it. You, Mr. Customer, decide which path you want to take. Cisco inside. We are doing it. We can supply components. We can supply white boxes and the full system. That is super important for the hyperscalers.
I think it is interesting we straddle this whole thing of software was always about flexibility, but then you sacrifice speed.
Hardware is great speed, but boy, the market's moving fast. You have to be careful what you invest in. This is a hardware speed with software flexibility. As you described it, what's important about this campus edition that we've announced? The campus edition, the smart switches for the campus, it's super interesting, right? It's where we're adding more security capabilities into the campus network with our smart switches. Those switches, there's a range of them, but they use our Silicon One technology as well.
Tell me, OK, I want to make sure we understand it. How do you pitch how this represents single architecture?
Is it because at the core, it allows us to refer to a single architecture which has these knock-on effects all the way through the management plane and how we interact, control plane, how we interact with it?
I mean, one way to think of it is like, what is a single architecture? What does it buy? Why is it important? It's important to us because it's cheaper for us to build these things because they're not different. It's important to our customers because they have feature commonality. In the olden days, it's like this class of devices behaved in a certain way, but they don't behave in the same way when you're in the wine closet. These are these fragmentation of feature sets. Another way to think of it is like x86. It's a one instruction set architecture.
You can run it in your data center server, or you can run it on your laptop, right? Now you have that the same program can run both races. It is a little bit the same way here.
I know when you guys first started, at least what it became, because there were many years before we knew it publicly, but prior to 2019. In 2019, of course, we had not had that ChatGPT Generative AI moment. As we look at Cisco's role in connecting disparate data, which is what we have always been good at, how does this help us stay in the lead in enabling customers to be able to?
I mean, it is a beautiful example, right? Because we did not have—there are a lot of things we did not have five years ago, even two years ago. ChatGPT is just two years old, right?
Yeah.
And it has changed everything. A nd there are feature requirements, especially in these large training data centers and networks, that you have to have much, much better load balancing because it's about your $10 billion worth of GPUs do not want to be sitting idle very long, right? If you drop the packets, then you have to restart and so forth. It is very important. There are certain features that were not invented when we built the chip that we then—our customer came in, said, "Can you do this?" We went back and we do-do-do. Yes, here it is.
That is flexibility that makes it.
That is powerful.
Yeah, so you're not worried about where things are going because it feels like there's a lot of pressure on you and your team to be able to say, "Predict the market," at the same time the market is changing faster than we can interpret things today as we're learning about it.
I feel very good on the path we're on. We have now launched essentially all our product lines. We have different letters in the alphabet for different classes. They're rolling out. We're working on subsequent generations and subsequent generations. I think we're starting to hit a stride here where we can continue to deliver value and innovation to our customers.
Final question. I was curious where this might go. What is the one thing that people tend to get maybe misunderstand and you think, "Well, it'd be better if everybody didn't have it"?
Maybe a misconception?
Yeah, it's a great one because we call it Silicon One. A lot of people think it's just one chip. We have one architecture, one chip, and it's a one-size-fits-all. That's not true. We purpose-built families for different roles in the network, for the service provider network, for the hyperscaler network, for the core router network, for the campus network. We have purpose-built devices for each of them optimized exactly for that role. The smallest device we have is less than 500 GB of switching capacity. The biggest one is this one right now. It's 51 TB. That is a huge range with or without tables and buffers. It's purpose-built. It's based on a common foundation, but it's purpose-built.
That's perfect. Martin Lund, Senior VP, Executive VP. I know you said I was going to do that. Executive VP.
Is that the way you say it?
It doesn't matter. You can call me Martin.
The hardware group. Thank you so much. It's always good. It's fun.
Oh, I love that. Thank you so much. Great job, Robb. And thanks so much to Martin Lund. He's just amazing. That brings us to the end of our Wednesday broadcast day. Amazing. It really is our honor and our joy to bring every moment, every message of Cisco Live 2025 direct to you wherever you are around the world, wherever you are tuning in from. On behalf of our entire brilliant Cisco TV broadcast and production team, all of your hosts, Robb, Michelle, Lauren, Z, myself, Steve Moulter, have a great night. We will meet you right back here tomorrow morning for closing day here in San Diego. Thanks so much for keeping us company on the live stream. Bye-bye, everyone.