Everybody, and thank you all for joining. I'm Matthew Smith, head of IR for Amdocs, and welcome to today's webinar. Today's plan is to take you on a deep dive to show you just how Amdocs is working to help telcos accelerate their adoption of generative AI using Amdocs aOS, which is the new agentic operating system that we've just launched at Mobile World Congress, that's purpose-built for the telco industry.
The main part of today's session will include a formal presentation, including some interactive demos to help things bring to life a little bit. After that, the lines will be open for a Q&A session with our sell-side analysts. For the broader audience, if you'd like to ask a question, you can do so over the chat feature in your web browser.
Before we get going, please note that today's session is designed to be educational and insightful. We won't be focusing on the financials or the guidance. That said, some of our comments today may be forward-looking and could include certain estimates and assumptions which are subject to risks and uncertainties as described in our filings with the SEC.
If you need more information on that, you can find it on the Amdocs Investor Relations website. To kick things off, let me introduce a familiar face for many of you, Anthony Goonetilleke, Group President, Technology and Head of Strategy for Amdocs.
Thank you, Matt. Great to have everyone online. I just wanted to introduce the group over here. We have Pilar sitting right next to me here, who heads our global consulting, who will be sharing her insight around our services space and how generative AI really impacts it.
On the other side, we have Liliana, who heads all of our strategic partnerships for me, and she will share how we work with our different partners to really drive the outcomes that really help our customers accelerate. With that, let me begin. As Matt mentioned, we will take you on a little bit of a journey through aOS. I think I've probably spoken to many of you over the last couple of months.
We did have a major launch at Mobile World Congress in Barcelona that was taken really well by all of our customers with many follow-ups scheduled after it. We'll talk a little bit about our outcomes 'cause at the end of the day, you know, this is something we're very focused on. Amdocs historically has really been an outcome-driven company.
If you think about what we do, it's been about production milestones, it's been about business outcomes, and it's really been about delivering the end business result as opposed to just another technology project. We'll share a little bit about the business results that we're starting to see trickle through some of these projects, and it's very exciting.
You know, when it comes to me and kind of my purview, I always tell our teams, like, at the end of the day, this is really what we should be focused on, because really this is what will move the needle. As I mentioned, we'll talk about our strategic partners. Now, Amdocs as a company has many partners we partner with all over the ecosystem.
These are companies we use internally within our software, companies we partner with to provide different parts of the system. There are several partners that really double down strategically. These are partners we work with all the way from thought leadership and thinking through initial strategic ideas with their R&D teams, all the way to go-to-market to all the way to actually launching out.
We had several of these partners at Mobile World Congress. We even presented at some of their booths at Mobile World Congress. There's really a good synergy between these strategic partners and us in terms of driving results. Finally, Matt and I will kind of host a Q&A. Keep all your questions for us, and we will set some time at the end of it to take any questions you have. Sound good, Matt?
Sounds very good.
All right. This, you know, for anyone that's new to the company, I'd just like to highlight this is just Amdocs by the numbers. You know, we're a $4.5 billion company with a very, very strong backlog in recurring revenue. As you can see, you know, we have a very, very high renewal rate on our managed services contracts. This really gets down to the point of our commitment to our customers.
You know, at the end of the day, you know, this industry is not one that has 10,000 or 50,000 customers. We are very, very loyal to our customers, and at the end of the day, that loyalty is returned, and so we're very focused on this. We have a very, very high success rate in terms of customer transformations.
If you look at kind of what we do on a day-to-day basis, this is also very critical. You know, a reminder to all of, you know, thousands of people that are developing software all around the world for Amdocs within my teams, and I always remind them, you know, at the end of the day, we are a company that is responsible for developing, creating mission-critical strategic software.
If we fail, basically, you know, I don't wanna kind of sound, you know, glorious, but if we fail, at the end of the day, society fails, right? If your phone doesn't work or data cannot be transferred, and you cannot access an application, or your enterprise cannot access an application, this is a fundamental failure, and this is not taken lightly.
If I look at some of the statistics here in terms of what we're delivering, you know, we are touching around just under two billion customer journeys daily from an Amdocs perspective around the globe. You know, if you look at our events that we're processing through our operations and managed services teams, you know, this is over 450 billion monthly usage events.
Not to mention, you know, around monetization. This is by the way, the 170 billion are daily rating events. These are data charging, rating, billing that go through the Amdocs systems, and these are not going away, right? If these, if you look at these events, if anything, they are only getting larger and more complex.
I think I've mentioned to several of you know, if you go back 10, 15 years and you look at my relationship with the service provider, I had one phone, and this is how I connected one phone, one eSIM, and I connected. Today, I have an Apple Watch, I have a phone, I have an iPad, I have a laptop that's connected, and now I have a little travel router that goes with me with a 5G eSIM.
Now suddenly you have five, essentially five eSIMs connected to my service provider. These events and these usage events and these billing events and customer journeys, we only see this trajectory go up. There is no sign of any of these coming down.
We feel that our job at Amdocs, at the end of the day, you know, many people go, "Okay, so what is your 30-second pitch about what Amdocs does?" We believe our job is really to take the friction out and to make sure that the customer experience we deliver to our customers is frictionless and seamless.
We know there's a lot of complexity in the back end. You know, there's a lot of hybrid networks. There are a lot of systems. There are a bunch of legacy systems that need to integrate. There are so many integration points. Our customers at the end of the day, don't care about this, so our job is really to abstract all of this. With that, let's jump straight into it.
When we think of the GenAI opportunity around the globe, I think I'm not sharing anything that would be new to you. This is mandatory. You know, we are at a fundamental place. You know, I've been in technology for a couple of decades now, and I think, you know, I always try to compare like, what were the two biggest events that really tectonically shifted technology.
I would think, you know, the introduction of the internet and IP and really it being embraced or generative AI. I would say, if I look at generative AI, the pace of adoption far outweighs the pace of the adoption of the internet, right? If you look at the launch of the internet and how, you know, browsers like Netscape came in and e-commerce came in, you're talking about 10 or 15 years before it became a household name.
You know, when it comes to generative AI, we are talking about a matter of a couple of years, and there isn't a person on the planet really that doesn't have access in using it. I think the enterprise adoption is mandatory. I don't think this is any surprise to anyone. I would say that the major shift really that we're seeing is a focus in terms of outcomes and ROI, in terms of what's being invested.
You cannot continue to invest, you know, very highly in infrastructure knowing that you have to do this, but you have to start showing outcomes and a lot of our customers are shifting their mindset into focusing on very clear outcomes to be delivered to the business. Another thing that's very important to our C-level technology suite is really having the right architecture.
Now, this doesn't mean that, you know, you will not use several different companies and large language models and ecosystem vendors and things like that, but it's very important to know what your North Star is and know how you'll govern this. Last but not least, one of the most important things is really around governance and organizational readiness. We are seeing now that, you know, technology is no longer the long pole in the tent anymore.
Organizational readiness, privacy when it comes to data, and governance are becoming really, really critical milestones. You know, Pilar sitting next to me and although we're not talking about it today, you know, I know you guys are spending a lot of work helping our customers kind of drive this organizational readiness and especially around data privacy and preparing the data.
Maybe that might be another session we need to talk about one of these days, Matt. On to the next slide. As we kinda think of this evolution, you know, this is not like Amdocs woke up one day and said, "Oh, you know, we're gonna be a GenAI native company." This has really been an evolution.
We have been very much focused on data and automation, and if you look at our strategic pillars at Amdocs, you know, this kind of predates the GenAI era. We even partnered with several companies, you know, back in 2016, 2017 when we saw these language models evolving, and really thought about how we incorporate it in our systems.
This is really, I would say a step function for us, but it's an evolution in terms of where we're focused on because we were very much focused on the data. You know, today you will hear a very common word that, you know, five years ago people didn't know what the definition was, but today people kinda drop it every 30 seconds. It's called ontology in the generative AI space.
You know, Amdocs introduced something called the Amdocs Logical Data Model, which was really one of the first frameworks of telecommunications ontology before anyone else did it, really connected all of the database entities, all of the business entities, your pricing, your policy, your customer entities to really give you a logical business view of how your business operates.
We've taken this and really evolved it to superpower, I would say, what we now call aOS. Just to kind of finish off on the history a little bit, we launched amAIz back in, I think, the middle of 2023, and this is really an evolution to that. As we kind of formed our...
You know, internally, we restructured ourselves as an organization because one of the things we realized was, you know, it's not just about meddling with things around the edges. You really had to reimagine what your organization looks like.
We kind of took everything we knew, and we said, "Okay, like, let's start with a blank sheet of paper and completely restructure the company, form an organization that really focused on this." It's a combination of product and services team that have been brought together. Here's where we kind of ended up, launching in March of this year, which was our aOS footprint. Now, this was kind of received very well from a market perspective.
Our customers kinda like what we've introduced and, you know, kind of one of my favorite pictures is down the bottom left-hand side there. You see Satya and Jensen on stage, talking about some companies that are doing some stuff. Then the bottom right-hand side corner is the logo of Amdocs.
You know, this is a very humble but proud moment, I think, for many of our employees to kinda see our logo on stage with these, you know, two great visionaries of our industry. Even as you look at aOS and you look at what some of our analysts are saying, you know, they are saying some kind of amazing things. If we just kinda jump to the next slide, Mike.
You know, Gartner kind of, I would say summarized it in a very clear way in terms of what we were trying to do. We were really not trying to fiddle around with it at the edges. We decided in the same way that when we built CES, you know, around 2016, 2017, we said we need to draw a line in the sand and build a cloud-native suite that is ready for the cloud on day one.
This is really what we're doing in the generative AI space. Launching aOS is our first step in going towards a GenAI native suite that can deliver those outcomes. And I think I've kind of covered most of these, I think. Let's jump to the next slide here.
I wanna spend a little bit of time talking about the framework of aOS and what it really encompasses. At the bottom, you kind of have our BSS and our OSS suite. These are still really our systems of record. These hold most of our customers' data around the world.
These hold the relationship, the policies, the procedures, the customer journeys, the business flows, how you do business. Every one of our customers around the world, believe it or not, does business differently, right? They may use our same systems, but they operate and go to market, very, very differently. I'll come to the middle layer because this is kind of at the heart of it in a second.
At the top is really an opportunity that we didn't play in, but we believe that, when it comes to the way the business operates and where the business functions, there is really an opportunity to use agentic capabilities in terms of our business and network flows. Also, we will talk about our agentic services, which is on the right-hand side here. This is, at the end of the day, really how we superpower our services, to be delivered under the agentic services banner, to really give you outcomes faster.
The last part that I wanna talk a little bit about here before we jump to the next slide is one of the things we allow our customers to do is, look, we are not coming to the market or not coming to the customers and telling our customers, "You know, dear ma'am or sir, you need to take everything from us.
It's a one-stop shop." We understand at the end of the day that this is going to be an agentic mesh framework, and there are going to be different companies, different large language models that our customers are gonna work with. We are allowing our customers to build on top of our ecosystem to really use our components to supercharge and superpower what they're doing. Think about, you know, a car being put together.
They don't have to reinvent the wheel. They don't have to go and reinvent the engine, right? They can take these key components and put it together. You know, one of my favorite analogies that you guys know is, you know, I always talk about the Lego store, right?
Where you go into the Lego store and you pick a bunch of blocks and, you know, at the end of the day, no one's asking a kid to go and build a Lego block. We are the people delivering those Lego blocks. You know, the imagination of our customers and the ingenuity of our customers are putting these together in the different shapes and forms. Let's kind of double-click on a couple of these layers, now.
The cognitive core, which really is at the heart of everything we do, this is the new layer that we've introduced, and this starts from a few different things. It brings all of the telco context. You think of Amdocs, you know, think of it as, you know, we go from zero to 100 in terms of super verticalized telco context. It's 40 years of delivering to customers, understanding the context, not just from.
You know, we're not a company that just does a customer management system or just does a workflow. You know, we're a company that takes an order, captures the order, handles the order, puts it in your billing system, sends it out to the warehouse so it can be provisioned.
Oh, by the way, provision it on the network, and don't forget to bill and charge and rate for it, and you may have to also go through a collections process. In terms of when you come to the verticalized knowledge of what we do, we have this deep understanding, and that is now all taken in and put into our Cognitive Core. When we're asking it a task, it understands all of this.
You know, I like sharing, you know, a small example just to highlight some of these. You know, you take something like Proration, for example, which sounds like a very simple thing, right? You need to prorate a customer's charge. You know, you take proration within a complex ecosystem, there are over 250 different database entities that has an impact on proration.
If you don't know this, you are highly likely to provide information that's not accurate. This is really, you know, key part of the differentiation. Let's stay on that slide for a sec, Mike. I spoke a little bit about the telco ontology. This is really where I think Amdocs shines.
We have been working on the ontology before it was called ontology, before people, you know, threw it around like hotcakes. This is really the Amdocs Logical Data Model. This is really how we know how a business operates to all the technical bits and bytes on the hundreds, you know.
There are some companies where we integrate to, you know, over 250 systems to provision an order and interfaces to provision an order. You need to understand this. You need to know how it connects in order to provide accurate results. You know, there are some people going, "Oh, like, what is the problem? Like, let's just take a bill and feed it into a large language model, and we can get it to explain it."
That's not what we do. We take context engineering to the next level. We take all of this information and the information we provide into kind of the non-deterministic angles of the system is next level. We believe the level of accuracy we get, the level of low latency we can provide.
'Cause remember, at the end of the day, we are still working with telecommunications companies, so latency is a super critical element that maybe can be tolerated in other industries that cannot be tolerated in our industry. These are super important things.
Then we took all of kind of the eTOM models, and we looked at what are all the roles across the entire telecommunications space, and we broke it into agents and sub-agents and tasks. We went right down to the granular atomic level to be able to make sure that, you know, whatever comes out tomorrow, we will have the components to be able to deliver it. Then as I go up the stack, you know, we communicate in very different ways, right? And by the way, this is also evolving.
Whether it's MCP today or A2A, agent-to-agent capabilities, or whether it's more traditional SDK API connectivity, we are also all over this. Meaning, we do it today and we provide today as the market evolves. You know, as we look at, you know, open core capabilities and some of our customers looking at how to use this, we integrate these back in, so you have these native connectors that you will be able to connect to our Cognitive Core.
This is really where I feel like we accelerate and differentiate because, you know, we're just not trying to, like, put your finger and, you know, stop the water from leaking. We are kinda reimagining what the telco experience could look like.
This is really why we thought, you know, aOS is the right name for it, 'cause we're really imagining how our customers could operate in the new world. I spoke about this. aOS is really just the next version, I would say, of evolution of amAIz, whereas amAIz really focused on specific customer care agents, things like that. aOS is really reimagining natively what your experience should be.
Now, with all of that talking, I'm gonna take a break and show you a demo here of how we operate and I'll step in here just to interrupt and highlight some things. What we are going to see is really a demonstration of what that end consumer experience could look like.
What you will see is a customer coming in and having a question or query, I would say, about their bill, and you will see how the agent responds. Now, I will tell you that we had this discussion 12 months ago. Many of our customers were looking at having a human in the middle in terms of interacting with this output coming in.
Our customers are getting very comfortable providing this input directly from an agent to the consumer. We are getting, you know, very, very high level of 90%+ accuracy, which, you know, most of our customers are very happy with. With that, let's kinda kick off the demonstration, and I'll step in here and explain a few things.
A new care business process starts as the system predicts a roaming overage bill shock scenario. A push notification is sent to the customers, instantiating a chat interaction via the CSP mobile app. The customer profile is retrieved to prepare and guide the interaction toward clear explanation and potential compensation. The general care agent leads the activity. Supporting sub-agents are activated to collect data usage data and financial info from the BSS.
What you see on the left-hand side is really an x-ray of what is happening behind the scenes on a
Root cause identified. The customer exceeded data roaming package. The bill explainer and charge explainer sub-agents provides the explanation. The sales domain agent takes over. Supporting sub-agents offer discovery and comparison. Sales could-
Can you just pause it there for a second? If you notice what happened here, you have an incoming call of a customer that's probably a little bit irate, got a bill shock. We give a very succinct, clear, empathetic response in terms of what occurred.
We also use that opportunity for a potential upsell. Now, we are able to do this at the speed of light because we're also looking at its entire history, its entire customer journey. We understand that the person is traveling. We understand that this could be a, you know, it's the right, I would say, the magic moment to interact with the customer journey.
To construct and present a quick resolution, a repurchase of same $10 roaming package. The customer negotiates a full refund of the exceeded $8 charge. As the requested refund is not within the policy limits, the billing dispute sub-agent evaluates the dispute parameters against business rules and authorization thresholds. Based on the customer's digital twin and the predictive analytics module, the billing dispute sub-agent calculates if retention investment aligns with policy.
So here-
The $5 waiver is ex-
If I can just explain for a few minutes. Here are a few very interesting things happen in the back end. You are looking at the company's policies and procedures. You're looking at a digital twin, and we're just zooming into the back-end systems here.
You're looking at a digital twin persona of the customer to look at the propensity of the likelihood for them to accept something or not accept something. You're basically offering them a discount or a voucher based on the policies and procedures of the company, of the mobile plan, and of the customer.
Accepted by the customer. The billing dispute sub-agent interacts with the billing system to perform the adjustment.
This is the automatic orchestration of it.
Data collected from the underlying care and billing is used to generate new insights, indicating meaningful retention value with an extended roaming package offering. When an upsale opportunity is confirmed, the agentic business and network workflows orchestrate the shift from care to commerce. The sales domain agent queries the catalog for suitable products, applicable promotions, and customer-specific entitlements.
None of this is guesswork, right? All of this is going down to the system of records, going down to your SOP, your standard operating procedures, looking at it. There's a governance structure in place, and the Cognitive Core provides all of these capabilities essentially back to the user. I think you kinda get the flow and how that works, and maybe let's go back in the interest of time, go back to the presentation.
I think, you know, summarizing this, you know, if you look at the time to value and you look at how quickly this can be done, this would generally be just to get the first response from someone you're speaking to on the phone. It's essentially a 12 to 14 minute call calling a call center.
You know, we can deliver these results within 45 seconds. It's very specific. It's a context of one to the customer, and our customers can integrate to the Cognitive Core at any level that they want, for example. I've spoken about CES.
We continue obviously to invest in CES 'cause this is super important in terms of delivering, running your business at the end of the day, right? This is the engine that continues to run your business. That is not going away. Now, the lines may change in terms of what is deterministic, what is non-deterministic. But in terms of making sure that you have a catalog with your offerings, making sure that your sales process runs the way it's meant to run, making sure that your customer care journeys are defined.
You know, monetization, and you can probably just go to the next slide, Mike. Making sure that these key fundamental functions are delivered within the policies and parameters, and making sure that, you know, your business can launch new offers and new products, and taking the friction out on how that's delivered is still a core part of CES. Still this becomes, you know, a cornerstone.
Now we're going to jump. Actually, one more before we jump into services. The agentic business and network flows is an interesting area for us. This is not an area we've traditionally played in, so this is really an opportunity for us, and we can just go to the next slide, Mike.
These are areas that our customers have where they run operations, and there's a huge labor kind of arbitrage there, right? There are a lot of companies that there's tens of thousands of people that do these functions. We believe that the technology available and some of our agentic capabilities that we essentially provide directly to consumers can really apply in expediting this, accelerating the outcomes, and even providing better outcomes in areas such as order to activation, billing operations.
This is really, I would say, a new opportunity for the company. In the same vein, if we do one more click and look at kind of some of these network workflows, you know, many of our customers are talking about how do you go towards a dark NOC. We are very focused on, you know, although we're not talking a lot about it today, we'll show you a demo at the end of some of this.
Think about service assurance. Think about having a closed loop system where, you know, a fault is being found, it's being addressed automatically, and a customer's being notified. This is really the future of tomorrow that, you know, generative AI can deliver, and again, really taking a lot of this labor arbitrage out of the equation. Next, I'm gonna hand it over to Pilar to talk about our agentic services and really how this is being super powered by aOS.
Thank you, Anthony. I'll take some time to ground us in what we mean by agentic services. Basically they're service domains that span across IT operations, data and AI, cloud, experience design, and quality engineering. These are activities that are transformed into intelligent, automated, and orchestrated workflows that go down to the level of the system of record, which is the unique thing I think here.
Each domain represents a concrete set of capabilities where we move from traditional, manually effort-driven work towards governed enterprise-grade workflows. What we can see, if we can move to the next one, I'll take you to how they operate in practice. Here, first off, this is a significant paradigm change for the industry. This is an evolution of what we have done.
Amdocs has always been very outcome-based, and we've seen that. This is us taking the services that we have done all along, and we are codifying that in a services as software paradigm, which is IP-based and defensible. We look at these domains and basically we see how the core components of the complex services, and you know, we do a lot of that.
We make it seem simple. You know, like you expect infrastructure to work. You expect your phone to work. When it's not there, it's disappointing. You don't want to see the complexity. This is taking it to the next level. What was once implicit expertise that lives in people's heads with 20, 40 years of experience serving our customers are now being codified in the agents and become orchestrated intelligence.
Now, that unlocks real value for the customers. Why? Because if you talk to almost every enterprise customer nowadays, legacy modernization is a key thing. They know they are keen to take advantage of GenAI, but they know in order to get there, legacy modernization is part of what needs to happen.
That's why I think agentic services are so exciting because they live in the lane where velocity happens with agentic adoption. We have structured processes. We have structured data. We know that is where you get ROI. This is for us part of where our differentiation lies. You know, we get to strategic outcomes faster because we have invested over the years, even before AI, in building the lanes that will get us there.
We're very committed to doing that in a way that's delivered at the pace of our customer readiness. Anthony mentioned, you know, customers are increasingly comfortable putting care agents talking directly to customers without human in the loop.
That's not necessarily the case in some of the other services, and we are committed to doing this at the pace of their readiness. Ultimately, what we have seen with GenAI is that as people get their organizations ready for it, they need to do it at a pace that makes sense for them. That will be varied according to different customers, different appetites, different regimes. We're a global company.
There will be very different data regimes, very different regulatory regimes, and we want to make sure that customers understand that we're there to basically give them as much or as little human in the loop as they need based on their current situation and their, you know, confidence and trust in the technology, and they will start putting workloads that they feel comfortable with and move on with that.
So if we go to the next slide, you know, you'll see exactly how this works in the demo. But given how important legacy modernization is, I thought I would take you to intelligent app modernization as an example. These slides basically takes you through the six phases that every customer going through intelligent app modernization would go to.
What you see, I think, and what we have seen over the years in this type of services, they're heavily constrained by the number of things. Typically, and especially in some areas like mainframe, heavily constrained by the talent pool. Very often, as a program leader when I was doing-
an aging workforce.
Yeah. An agent workforce and low-
Aging workforce. Yeah
... aging workforce. That's aging.
Yeah.
That's been the case I think since I've been in technology.
Yeah.
Like we had, you know, COBOL developers.
Yeah
In our backlog many years ago. I know, I'm not the only program director that has done this. I have shaped mainframe modernizations or migrations based on the available talent. That was my key constraint.
Mm.
Now this is changing, and it's changing in very exciting ways because it's opening up opportunities that we didn't have before. What I'm going to take you through is I think, if we go through the journey, what you're going to see in the next slide please, is how the collaboration between humans and agents can happen.
What we show here, and I'm going to use the example of workload discovery because I think it's the best one. We have systems of agents. This is unique in two senses. The first one is that these are not generic bots. They're not accelerators that are put forward by a platform.
They are agents that benefit from both our collaboration in R&D with our partners, which is very deep, and the expertise, the human expertise that has all of that implicit knowledge and all of that explicit data knowledge and structure at the system of record level.
That's all baked in and encoded into our agents which actually operate as a system and in an integrated fashion, going from task to sub-agent to agent in a coordinated way. What this allows us to do is do what I think is the ideal situation, the best of both worlds. Agents are able to do volume at industrial scale. They can do that consistently. Like, no human can read, like, lines of code the way agents can do, and they can do that in parallel.
Humans are able to do what we do really well, which is understand the change in context, the parameters of the organization, the risk appetite. We live in very interesting geopolitical times, so your main modernization planning might make sense based on the readings of the past, but that's where the human comes in and goes, "Well, actually, this just happened, therefore, we need to refactor the modernization recommendations."
That allows you to spend time doing scenario planning, for instance, which, you know, was costly and expensive to do before. I always tell people that workload discovery is where migrations slow down. You start, and very quickly you find that you're not going to go to the level of depth that you would like to do if you do it manually.
This is actually the risk in that for our clients, and it will allow us to focus on where the energy is. I'll take you to the demo quickly. I'm going to save myself some words. I'll take Anthony's example. What I want you to focus on here is three things I think that are super important for the agentic services.
The first one is that. I'm taking you to the example of intelligent app modernization because I want you to see the depth and the thoroughness of the agents. We're going to focus on three areas here. The first one is the business outcomes part of it. This is pre-configurable. For this type of services, it matters a lot because each of our customers is different.
If your app modernization target is to do a data center exit, that's time. Time is going to be more important to you than maybe other considerations. In this example, it's going to be time to market, but it could be any of the business considerations that our customers have, and that's how intent gets transformed into outcomes, and it's very important for us to track.
I'm going to call attention to the left-hand side where you have the level of depth of the agents and how they navigate all the way down to the system of record, which is very important. Finally, this is the difference from what Anthony showed you before. Anthony was showing us the outcome as it's as the end consumer sees that. This is very different. This is the end user experience.
It's the enterprise user view. That's that pane there is where humans and agents collaborate, where you will have, or you can see how we get reports. We get artifacts from the agent that then the human will evaluate, stress, assess, and then decide, "Okay. Yeah. Go, go for it.
Let's have a look at it.
Let's have a look at it.
The product manager flags campaign management as the root cause of campaign convergent rate. The orchestrator proposes a modernization path, starting with cloud migration and quality engineering to control risk. The dependency discovery agent confirms the constraint is isolated within the campaign engine with no impact on surrounding systems.
Based on that analysis, the migration planning agent recommends a targeted refactor to improve segmentation and cross-channel relevance safely and within scope. The refactoring blueprint agent pinpoints minimal changes to remove tight coupling, enabling scalable segmentation and consistent activation. Campaign behavior stays intact, and with PM approval, refactoring begins.
Okay, let's stop here because this is the important moment. This is the human-in-the-loop moment, and that's where the appetites of clients are, is going to vary. You would spend time here interrogating the output that you get and, you know, having the organizational conversation.
And I expect that there's going to be a range of how much or how little our clients need there, and we're seeing that already in engagements, and we're happy to accommodate their pace and their comfort because ultimately, outcomes are built on trust. In this type of work, trust is built with a human. If we can move on, I think you're going to see both-
The refactoring blueprint agent pinpoints minimal changes to remove tight coupling, enabling scalable segmentation and consistent activation. Campaign behavior stays intact, and with PM approval, refactoring begins. The resource provisioning agent validates the modernized campaign services in a secure sandbox under peak load.
The deployment agent then migrates them to the cloud with repeatable deployments and built-in safeguards which are ready for controlled validation and staged rollout with minimal risk. The quality engineering agents validate the modernized campaign foundation end to end, covering segmentation, eligibility, and cross-channel activation.
After execution, the governance agent confirms readiness is green, and the release is approved for deployment. The product manager shifts focus to usability to improve campaign setup. The orchestrator activates experience agents to drive adoption. The customer journey agent identifies where teams struggle with segmentation and eligibility, and the opportunity builder agent translates these insights into a prioritized UI improvement roadmap. The UI designer-
Mike, if we can stop there because I think we have seen the gist of it. What I find is incredibly exciting is how tightly integrated all of the different agents are, and that they are very specific to the industry. They're, you know, they really understand, you know, situations. If you're doing a migration when a user's trying to buy an iPhone, well, revise that, you know.
Yeah.
This ties back to the policy as code we were talking about earlier, which I think really matters to get to outcomes fast in the industry.
Right.
I'll give that back to you.
Okay. Thank you very much. It's exciting. We're gonna very quickly rush through some of these slides, just given the timing, but we wanna talk a little bit about the outcome here. Let's go to the next slide, Mike. You know, we always, everyone knows about the traditional bots.
Then we had kind of the, I would say, the AI-enhanced era, and now we are really at the agentic era, where really it's a context of one, where you're having a digital twin that knows the propensity that someone will take an offer or not take an offer or how they will react. You intuitively look at the intonation on how they communicate, and you react your response or change your response based on that. That is really where the era that we're in.
With that, I wanna share some of the outcomes in terms of what we're seeing. You know, you see things like, you know, this is one of our customers that deployed it via WhatsApp channel. You know, 135% NPS improvement. Their CSAT increased. You know, something we didn't think, faster payments.
Why faster payments? Because they understand the issue better. They don't have any complaints, so they pay it and move on with their life, right? 40% reduced wait times. These are really some of the outcome metrics that we love to look at and love to track. If we jump to the next slide, this is one of our tier one North American providers.
You know, when you look at call centers, at the end of the day, there are three primary metrics everyone always measures: AHT, FCR, and NPS. Average handling time, first call resolution, and was the customer happy? To see all of these three things increase with a customer is, you know, being in the industry for so long, this is something that's pretty cool.
But I also don't wanna forget that last point at the end here. You know, from the time we started to the time we end, we ended up with 60% less tokens as we tune it. We understand that, you know, it can't just be cost rising just to get a good result. We also have to work on the overall cost structure of what we're delivering. Let's jump to the next slide, Mike.
You know, we've spoken to many of you guys about different POCs, but we have many customers now using our data and kind of AI capabilities in different shapes or forms. You know, we think we're delivering something that either our customers are looking at or already trying to build or needing to build desperately.
Really, it's a place where we feel that we're meeting them. Now I wanna introduce Liliana here to talk about our strategic partners, which really is a core part of our strategy 'cause we believe these four partners really accelerate everything we're doing, and we work with them very deeply. Liliana.
Yeah. Thank you, Anthony. Very excited to be here to talk about those four strategic partners. Amdocs has a wide variety, as Anthony mentioned earlier, of ecosystem of partners, but right now we're going to focus on those big four: AWS, Microsoft, Google, and NVIDIA. What is so special about this partnership, the strategic partnership that Amdocs is having with them?
First, we are sharing vision. Combining Amdocs' deep telco expertise and offering that actually we are managing it for what? Four decades right now. Combine it with the fastest technology that those amazing companies provide us. We wanna lead the telco industry together into the agentic era by offering the different unique cross-domain solutions that were presented throughout AOS and the different components of AOS. Why those strategic importance are coming to us with those partners.
As we said, it's the deep knowledge, but our partnership is very intimate and very multidimensional. With those four partners, we're going deep as R&D to R&D. We're developing engineering to engineering solutions. We have a shared go-to-market. We have a shared sales campaign, shared offering, and that actually opens up the market share that we have together.
We co-innovate more and better together with all the different amazing things that Anthony presented here today, as well as we are using them internally. Even Amdocs is transforming and becoming more agentic using some of those AI transformation with our products and our services. Which are the domains we're focusing? Obviously those companies are broad and we are trying to focus on domains that might be relevant for the telecommunication industry.
When we're looking at the different, you know, domains that when this 1+1+4 here becomes 11, is how we are extending this aOS suite into more agentic power customer experience into agentic like we discussed right now, the services, right? We are running together into modernization, different levels of modernization, and so on and so forth.
Now let's see how we're translating those partnership into live examples of where they're embedded within aOS value proposition. I'll just share a few examples. One of them, and we've seen a demo just before, is how we're embedding our Cognitive Core within the telco contact center that right now we see a lot of dominant solutions such as Gemini Enterprise for CX for customer experience, as well as AWS the contact center.
It's all integrating with Amdocs telco specific knowledge that provides, as Anthony described before, way more accurate, and trustful results, right? Using the amazing tools that we are providing with our partners. If we're looking into another example, maybe it's a flagship of, an example that shows how deep we are going in our partnership, and this one goes with what we call Customer Engagement Platform.
It's a product to product integration between Microsoft Dynamics 365 that has agentic solution for the entire customer experience from marketing to sales to customer care to CRM to case management. All of that is actually pre-integrated with Amdocs BSS Suite mainly around the care, and, commerce as well as CPQ.
Those solutions provide end-to-end new customer experience with the agent and agentic solution inside to provide new customer experience through B2B and B2C offerings. We talked a lot about the agentic services. Maybe just to point out, obviously what we have heard today is strongly embedded within additional tools that we use, like for example, the Amazon Q Developer Transform, right?
We use it and embed it when we are actually creating our own tools. Same goes with Microsoft Fabric and so many more of the latest tools embedding within our different services. I'll touch quickly on two more examples, and then we can actually see this demo later on around autonomous network, right?
As Anthony mentioned, we are partnering here with NVIDIA, doing digital twin of the network. We're actually mapping that through the RAN network and trying to predict some faults and actually to prevent them. We can simulate that, we can test that, we can deploy that ahead of time before it goes into the production.
So all this preventive care and service assurance and network assurance can be done with our partners here. The last point here on the shared offerings around Sovereign AI, we are partnering obviously with AWS. They have a very robust Sovereign AI. We are the service arm and sometimes the monetization arm for them as well.
As well as AI Factory, we see that as one of the most important use cases right now is how we are actually servicing our telco customers with their potential enterprise customer who wants to use the AI infrastructure. That's where we're bringing all the stories together. Now, I'll finish this in one sentence here. Amdocs is doing it internally as well. We are transforming, Amdocs itself is transforming to become more a frontier firm, and we're doing that hand-in-hand with a very strong collaboration with Microsoft.
Microsoft has this vision of very advanced companies that can transform theirself through the agentic phases, and Amdocs going through this phase with Microsoft, and we are looking at embedding more AI and automation through copilots, through different tools we have out there, through our different business units within Amdocs, like sales, like finance, IT and so on and so forth.
Now I'll just give a glance of few real examples coming from our customers. One case study we see here is Vodafone. Anthony mentioned how deeply we understand the data and data structure this in the telecommunication industry.
Combining a Vodafone request to actually migrate very complex enterprise data warehouse into Google Cloud, Amdocs is the one who actually performed this work and moved everything into the Google BigQuery, and that resulted in much faster time to insight. It allows them to have AI native data foundation that can drive way more use cases, and obviously it reduced some of the cost of running those data platforms.
I'll give another example, and it's similar to some of the examples Anthony shared before, and it's an APAC tier one customer that we implemented the CP that I presented before. It's the customer engagement platform that, as I said, it's Dynamics 365 of Microsoft and Amdocs Commerce and Care combined together. Here in this case, we modernized the customer service and unified all the customer channels into one platform.
It's amazing to see like what we've seen with Anthony presenting some results with another customer and this customer. We are improving drastically. One of the, I think, high ROI outcome is the customer experience. Here in this case, the case management was improved as well as their NPS that we've seen with the other customers and first call resolution. Overall, those agents, over 4,000 agents could provide services to their you know large customer base. Then there is another example. Anthony, you wanna share with us?
Yeah, sure. I think we're actually gonna show a demonstration here. Matt, we might actually not show the entire demonstration. I might actually just talk about it given the time, 'cause I wanna leave room for some questions.
Absolutely. We can put this on the website as well, so the people can see it.
The demonstration we were gonna show was actually our partnership with NVIDIA and Omniverse, and how we basically created an entire digital twin of the network and allowed autonomous healing problem detection by using a digital twin.
In the same way, you know, we think this capability is relevant not just to create a digital twin of a customer, but to create a digital twin of towers, to create a digital twin of the network, and really it's very pervasive in terms of how we can use it. With that, Matt, maybe we'll just kinda pause for a second and open it up for any questions that people have.
Absolutely. Let's go.
Jill, do you have any analysts on the line?
Thank you, ladies and gentlemen. If you have a question or a comment at this time, please press star one one on your telephone. We'll pause for a moment while we compile our Q&A roster. Our first question comes from Tal Liani with Bank of America. Your line is open.
Hey, guys. You actually have, Tomer Zilberman on the call.
Hey, Tom.
Anthony, maybe two for you.
Sure.
Hey, guys. Wanted to ask, first, you talked about amAIz, the amAIz platform earlier.
Mm-hmm.
Is this aOS, is it an extension of the amAIz platform or is it a replacement of the amAIz platform? The follow-up to that is, you know, can you give us any, you know, commentary around what the traction was or is for amAIz and how aOS changes the adoption rate in terms of agentic, because you had some earlier key commentary around customer readiness for agentic. Does that change the migration path? Does that speed things up? Does that slow it down, given where your customers are in terms of being ready for agentic?
Yeah, that's a great question. Think of amAIz as kind of our version one entrée into this. What it provided was really agentic capabilities. Think of aOS and the Cognitive Core as a complete re-imagination. Yes, it incorporated every single thing we learned from amAIz into it, but it basically allows us to the second part of your question.
It allows us to meet our customers wherever they are at. If amAIz kind of provided a certain functionality for certain set of customers, aOS now allows us to help our customers build out their agentic fabric to help accelerate their transformation, to help in a mainframe modernization.
It gives us, as a company, a lot of different, I would say handshake points or insertion points with our customers, to be able to provide solutions at different places. This is one element. The second element is really just bringing it all together rather than having bits and pieces, so our customers can pick and choose how they integrate with us.
We feel that, you know, we've been thinking about it obviously for probably, I would say the last 18 months or so, and aOS was a framework to allow our customers to run faster, to integrate at wherever they are in terms of their customer journey and their build-out of generative AI architecture.
The third thing was really to focus on business outcomes. I mean, we doubled down on this like we probably never had before as a company and said, "You're not building something unless you're going to really deliver a business outcome at the end of the day.
Got it. Maybe as a follow-up just to that last point, are you thinking about any changes in your pricing model for amAIz versus aOS?
Yeah, it's a great question. Look, we are testing the waters. We're thinking about it. As you can imagine, I don't think anyone in the industry has reached any level of equilibrium in terms of what monetization looks like. I think, you know, everyone is going, "Well, how do we charge for it?
Is it tokenized? Is it not?" But I'll tell you one part about it. Some of the savings that these things can be delivered are pretty big. Definitely we're thinking about, you know, what are the different monetization models that can be out there that can basically help us, but we're in the middle of that process at the moment.
Got it. Thank you.
Thanks, Tom.
Ladies and gentlemen who are listening via the webcast, you can also submit questions by using the chat function. One moment for our next question. Our next question comes from Timothy Horan with Oppenheimer. Your line is open.
Thanks, guys. A great presentation. Anthony, obviously the AI adoption by the carriers can massively improve productivity and quality. You know, where do you think? I would assume they're under a lot of pressure to do so or will be competitively. You know, when do you think adoption really starts to accelerate and what's kind of the main barrier to adoption?
Hey, Tim. It's a great question. Look, I think in terms of adoption as a technology, everyone is there, right? Like I don't think, you know, I've met hundreds of customers at Mobile World Congress from around the globe. I don't think there is one customer that isn't doing anything with it or trying to accelerate their customer journey. I think where we are right now is really saying, "Okay, so we've got this in place.
How do we focus on the outcome that can be delivered?" That's really, I would say, where the focus is right now. The second part, I would say, is trying to figure out what that blueprint looks like. 'Cause you need some resemblance of a blueprint. If not, it just becomes a Wild Wild West, right?
'Cause, you know, everyone is like, "Yeah, I'm doing AI, I'm doing AI." And so you can't add 1,000 different vendors doing, you know, 7,000 different things into the equation. So it's about determining who your domain partners are that you want to partner, and ones that would really commit to delivering an outcome on this, I would say. If you want to add anything on that. No. Good.
Just on the digital twin, I think you're implying that we're gonna have digital twins for every piece of their business, not just the network.
Yep
Which would seem to be a game changer to have a digital twin for the network, but also for the entire IT stack and maybe even, you know, customer databases and on and on and on. I mean, where are you with implementing digital twins?
Yeah
Where are the first places which you see it, yeah?
Yeah. I'm a huge fan of the NVIDIA Omniverse framework and architecture because I think really that unlocks so many different things. We are focusing on a few different spaces where we're starting to get a lot of traction with customers. Definitely in the network space on specific segments.
When it comes to network rollout, when it comes to service assurance in terms of problem-solving, where you can kind of do spectrum optimization, where you can do, you know, rerouting of networks, I think digital twins can play a very, very big part there in terms of problem resolution. The second part, which I think, like, probably got more interest than I thought it would have gotten, was when we presented our view of the digital twin persona of a customer. We train...
Basically, we take all of this customer information and data and propensity to do X, Y, Z and put it into a certain persona. When you're dealing with a customer, you don't have to guess. You know, you're already going into it going, "Hey, like, I have this marketing offer. There's a 75% propensity that a customer with this persona will actually lead to a conversion in a sale."
Right? That part is very, very interesting. As you can imagine, our telcos sit on a lot of proprietary data, and they can use it internally however they want to sell their own products, right? There's no privacy issue or anything there. I think building those personas and using those personas is a little bit of a game changer versus, "Hey, here's a technology.
Let me hope that this technology delivers the business outcome that I want." We're pretty excited about this, and we're really starting to incorporate it in many areas of the business.
Thank you.
One moment for our next question. Our next question comes from Shlomo Rosenbaum with Stifel. Your line is open.
Hi. Thank you very much. Would you say that the aOS is geared more towards small and mid-sized clients, larger clients? Who is most likely to be adopting this? Would it be the ones that need more help getting, you know, usage or in implementing GenAI, or is it gonna be the larger customers that are already more sophisticated over there?
Thank you, Shlomo. I love the question because that's something I forgot to mention. One of the fundamental principles of aOS is that we have different customers at different stages. If you take our tier zero or tier one customers, you know, they are already way down the journey, right? They have an ecosystem, either they're building it internally or not.
With aOS, like, we had a meeting this morning with one of our large strategic customers, and I was in the meeting, and one of the discussion points we had was where they can integrate. They wanted to integrate at a couple of levels down to compare to what maybe a tier three or tier four customer would discuss with us, right? We had some smaller customers going, "Hey, that's fantastic.
Let's take your entire agent, put the entire aOS infrastructure in and run with it." Whereas some of the bigger ones may wanna integrate at different points. At the end of the day, you know, we wanna, like, you know, like, kind of like that no customer left behind comment, right? Like, we wanna make sure that aOS can integrate at different places, and that's why by design, we also made it very open.
If you think of the different layers that it comes with, it comes with the fully fledged agents. It goes to the super agents, it goes to the sub-agents, it goes to the tasks, it goes to MCP integration, agent-to-agent integration, and all of these are very, very different integration points that different tiers of customers will come in, and they are, by the way. We see it even right now.
By the way, one of the challenges that, you know, we tried to overcome with this is that no two customers have the same journey or the same experience or even the same destination that they wanna get with generative AI. It's not like, you know, there's a cookie cutter shape, and this is really what AOS is also trying to address. Thank you for the question, by the way.
Do you envision this pulling along modernization ahead of clients going ahead and implementing aOS? In other words, when I think about the impact on the company, are you gonna go ahead and say, "Hey, we've dangled this in front of you, and therefore we have to do X, Y, Z amount of work for the next two years, and therefore you're gonna have this"? Or is this something where you're gonna start out earlier on with the clients? Just maybe give us a little bit of color on that.
Yeah, sure. You know, at a macro level, you can think of software really broken into two parts, right? Part A is what we call non-deterministic. This is essentially 100% the generative AI space, right? There's the deterministics, which are the system of record, which is the standard operating procedures, the catalogs, the policies, the pricing, the billing, things like that, the deterministic part, no matter what anyone says, is not going away tomorrow.
Now, the lines may change, the lines may be blurred, but at the end of the day, you still need to collect an order, deliver an order, provision an order, operate, bill, collections, do all of those fundamental core mission-critical pieces while delivering all on all of these non-deterministic promises. Yes, we believe that some components will be pulled in, right?
I'll give you an example with one of our customers where they're taking aOS, and they're like, "Well, in order to get the maximum out of generative AI, we also need a modern enterprise catalog." Right? They will go down the journey of modernizing the enterprise catalog because what they have now, for example, it's an older version of a catalog, which takes three days to update a product, right?
So you can't really do some of the capabilities like digital twins and propensity to buy-sell, which are all real-time events. You cannot necessarily do that with legacy systems. You know, you have this ongoing argument, right? I mean, when we look at a macro perspective, do enterprise systems stay, do they get replaced? Things like that.
We believe that there is a world where, yes, there will be deterministic components and non-deterministic, and in our world, we have to balance these two, and yes, I believe one will also pull the other.
Maybe next question.
Okay, and then to...
Yeah. Go on, Josh.
Go ahead, Sean.
I just had one last question, and I'll hand it off to someone else. Just in terms of the example that was brought with the customer service, that was being shown earlier, the bill explainer came out a while ago, and you know, maybe you could talk about the adoption on that and whether that would be a good indication as to the adoption on AOS or is that not something that I should be looking at in terms of you know, how one is indicative of clients' propensity to adopt?
Yeah. So, the bill explainer and the customer care was really, I would say, the cornerstone use case that came out of amAIz, right? Now, obviously, that's expanded and, you know, we are now cross-selling and upselling and doing all sorts of things.
But you're right. I mean, that's essentially where it started from, and we've gotten very good kind of customer traction and adoption where customers even wanna integrate at different levels of the tasks or sub-agents to help them. For example, they might say, "Actually, we do have a, you know, a care GenAI agent, but we really like the level of accuracy you guys give in A, B, C, D domain. Can you incorporate it and integrate it?" That is happening at the moment.
When we look at aOS, that then expands it to, number one, different domains, number two, different insertion points and integration points, and number three, kinda what you mentioned before, it kinda pulls in some of the capabilities of the underlying system. It also creates an opportunity for modernization on some of those components.
Like billing, right? I mean, think of monetization. We have some launches, you know, that happened lately about very, very different monetization mechanisms. Even as kind of 5G standalone and 5G advanced starts to evolve, people are looking at very, very different ways to monetize it.
Now, these capabilities still need to be built into the enterprise systems in order to be exposed. Yeah, that's kind of the, I would say, the evolution from amAIz to aOS. aOS just kinda expands that and gives you more depth.
Okay, thank you.
I think we've got time for.
I'm not showing.
Oh, sorry.
I'm not showing any further questions on the phone line.
Okay. I think we're well over time. This has been a fascinating discussion. I wanna thank our presenters for today. You did a wonderful job, Anthony, Pilar, and Liliana. Awesome. If you have any questions for us, and we didn't get a chance to address your question on today's call, by all means, call out to us here in the Investor Relations department. We'll be pleased to get back to you. With that, thank you for your time and your interest, and we'll wrap it up.
Thank you very much.
Thank you.
Thank you.
Thank you.
Ladies and gentlemen, this concludes today's presentation. We thank you for your participation. You'll be now disconnected. Have a wonderful day.