Don, thanks for joining us.
Of course, yeah.
Looking forward to our conversation. There's just so much going on in the industry at the moment that it's good to have, like, a more technical person here to kind of do this.
Excellent.
Maybe if you start off, like, talk a little bit about your background a nd then we kind of go into the presentation.
Excellent. So I've been with Pega for 26 years, and my background originally started with Pega in our support organization and working directly with clients like Citibank, Goldman Sachs, on implementing workflow automation solutions, integrating them, a lot of the work that I did, right? Because for workflow to really work, it's got to integrate very tightly into the existing apps the clients have. And I took a lot of that implementation work, moved into our engineering organization. You know, got, frankly, pretty bored at sitting in a cube writing code all day. Worked myself back into the client-facing side of the organization, so as like a solution architect.
Yeah.
Eventually led our field [roles with engineering and architecture function, which eventually just sort of just turned itself into getting the CTO office. Spending a lot of my time, probably about 50% of our time, products, and 50% [inaudible]
Yeah. Okay, that's interesting. And then talk a little bit, like, did you say 26 years?
Yeah.
W ow. Okay. Congrats.
Yeah.
Talk a little bit about the evolution of Pega from a product perspective over those years. Because that was like, was Windows out there already, or?
Yeah. So I joined the company when we had transitioned into kind of thin client, Windows-based- kind of development environment. The first set of thin client, browser-based, end user apps. Pega had always had a kind of metadata and, I would say today we would describe it as very JSON-ish tagged way of representing data, which meant we fit very well into HTML architectures.
Yeah.
We fit very well into XML and now we fit very well into JSON-based architectures, which is where everything is going. But like, over my years, a couple of things have evolved and a couple of things have stayed the same, right? At origin, Pega has always been, can we automate workflows and decisions at scale for pretty manageable users that have high degrees of success? So they usually are global. They have multiple business lines, which means they need to work in different regulatory environments. They have multiple products, and they want to be able to have not a bunch of siloed systems.
Yeah
But actually one common system and way of doing work across all of them. That's always been at the core of what we do. That means automating business rules and decisions as well as the workflow. Over the last many years, there have been big evolutions. One is in the architecture, so a big shift into the cloud, which was both a business shift, and I'm not going to try to talk about our recurring revenue sort of transformation, ask Peter and our finances about that. But that's also been a big shift around moving our client workloads onto Pega Cloud, which is our SaaS offering, our hosted offering. And then the other one, especially over the last 10 to 12 years, has been the impact of AI.
So we'd always been in what I would call the AI space. We used to call AI expert systems- back when it was rules-driven. We acquired a company called Chordiant, I think, about 15 years ago, and they brought predictive and adaptive analytical AI to the equation.
Yeah.
And we've been using that with clients to drive massive decisioning scale, so really getting into the front of the client conversation and drive decisions that actually drive revenue across the upsell, retention-driven decisions. But now taking that same analytical capability and pulling it back into the process side. So process improvement, suggestive ways to make process improvement. I think that sets us up really well right now for this Gen AI moment, because we've already got the architecture and the books and the datasets in place to really leverage AI in a pretty effective way.
If you think about it, you guys put out the vision all around the Autonomous Enterprise. Like, what that sounds like a very firm. Can you just pull that for us a little bit?
Yeah. So, I think one of the things that I would do is sometimes it's easier to take the concept of autonomous- and replace it with the idea of self-optimizing. Because, you know, I don't think, I don't think anybody wants to turn a business into a completely self-running. Leaders, stakeholders- Still want to be able to make decisions. But what I think they want to be able to do is provide guidance into their technology. I want to optimize customer engagement to drive these problems. I want to optimize this process to reduce cost of service. I want to optimize this process to maximize throughput.
Being able to set those goals into the system, and then have the system, through a combination of automation, Analytical AI, and Generative AI, automatically adjust the process. To deliver those business results. That's the vision that I think of when I talk to folks like COOs and heads of customer experience, that they're really excited about. And I think with Gen AI, it's a vision that we're not that far from.
And then if you think about it, then that's almost like the Holy Grail, a little bit. You know, that would be good, if you can do that. H ow is Pega kind of well-positioned to or why do you think you are well positioned to deliver that?
I think it's important to understand that clients are going to go on a journey if they go there, right? And, you know, I got to sit in Bill McDermott session, and one of the stats that he dropped had to do with people's voluntary, right? And we actually did a study because we actually have a lot of data about how clients use our software, so we actually know when people are. We actually see it something like 90-day. D ouble checking across different applications to get something done. Right? So there is sort of this first maturity level that organizations need to go through of just automating the simple tasks.
In some cases, it's just removing friction, it's removing points that are, that are disconnected. By doing that, you put in a very simple kind of case and process structure to capture that and manage that. But now what you're also doing is you're actually building data history. Right? So you're building history of how the work is getting done. So I look at an organization like Citibank. We process credit card investigations. Well, in their Pega environments, they have tens of millions of records of how that work has been processed in the past. That is the secret sauce then to drive the analytics and the AI.
Yeah.
To find where are the optimization capabilities? How can I use the past to predict what happens the next time we get one of these? And how can I proactively prevent bad things from happening or proactively optimize this process going forward?
Yeah. Okay, it makes sense. And then the other part of that, like extending that a little bit, like, if you look at your flagship software suite, it's Pega Infinity, and it's like now in version 23. Like, this Constellation, like, think about Constellation.
Yeah, Constellation is user experience. So, you know, one of the things that I think we've all seen is the desire, even in our internal lives to have consumer-grade user experiences. So that means I want things that are fast, I want them responsive. And, without offending any developers here, if you ask a software developer to build a user experience, you probably aren't going to get a very good one.
Yeah.
Right? It takes lots of investments from designers and usability thinkers. And what we wanted to build was something that anybody who's designing a process, be that a process developer or even a citizen developer to use low-code term. If they laid out, "Hey, I need to collect these five pieces of information at this step of the process," they would get a great user experience for doing it. And that user experience would be consistent on a web app, a mobile app.
It would provide APIs to allow them to plug it into their existing websites. It could plug into their existing Salesforce system, their existing Dynamics system, and do all of that seamlessly. So Constellation is an architecture that allows me to create a very simple set of controls, gives the information I need to collect at this step of the process, automatically generate a world-class user interface that both works across all those channels, but also easily integrates the other products.
Yeah. Okay, that's interesting. And then, you mentioned AI quite a few times already. Like, if you think about it, you have already like a good few kind of solution sets out there. How do you think about AI in the context of, like, Pegasystems and where you kind of then fit into the industry?
Yeah, I think it's important when I talk to our clients, like our CIOs and chief analytics officers who are running kind of AI practices for their clients. It's important to think and understand sort of the nuance of AI. Right? Because it's very easy, we love generative AI models, and they're very cool and very powerful, but large language models are actually a subset of the broader capability that we have inside machine learning.
Yeah.
Right? And if you ask GPT-4 to simple math, it's not very good. It actually gets it wrong a lot of the time. Right? And even if it did get it right a lot of the time, it's a really expensive calculator.
Yeah.
Right?
Yeah.
So, what I want to be able to do is have different types of AI, some of which are large language models, and that's really good at creating things, right? Generating information, generating a new workflow, automating the integration mapping between my workflow environment and my four legacy systems So we've got about 20 or so different capabilities in the product that we've released to do that. But—and you can think of that kind of as like right-brain data.
Yeah.
It's creative AI, right? It makes stuff. There's also left-brain AI. It's AI that is analytical, it makes decisions, it looks at numbers. It predicts what's the churn risk of a customer. It predicts what's the potential value of making this offer to a customer. It predicts how likely is this KYC process I'm running with my commercial banking customer likely to run into a regulatory issue?
Yeah.
Right? And the important thing about that analytical AI is, one, it's predictable, it's explainable. I need to be able to tell my client and my regulator how I make decisions, and I don't need 40 billion records. I can train it on a couple hundred thousand records. So what we want to ensure our clients do is use analytical AI for those kinds of use cases, next best action, process optimization, and then use gen AI to accelerate the way in which they could deploy their cases in Pega and connect across the different technologies here for a business person to actually drive optimization into their business process without having to be a developer.
So then, so if I understand you correctly, so that would be the Analytical AI is the stuff that we had before already in a way.
Yeah.
Like, you know, Gen AI was this year, but the analytical AI you should have worked on for a while, or?
The analytical AI has been out there for a while. What I think is interesting about it is the Generative AI boat is lifting all boats. Because now all of a sudden, across the C-suite.
Yeah
People are suddenly aware of what AI can do.
Yeah .
Aware of the value that it can deliver and the fact that it's real. So as we talk to clients, they're happy as long as they can drive business value, whether it's practical AI, generative AI, the most important thing to my client is: Does it improve my customer experience? Can I tap attach driving revenue, and can I attach it to improving the efficiency of when my humans get work?
And then the Generative AI, like, more to help people to interact with the system, like, if I understand you correctly a little bit better. D o I think about that as, like, a copilot idea, or like, how should we do that?
I think you can think of it as a copilot. We've also been using the term autopilot.
Oh, yeah.
The reason I like the term autopilot is I actually want my Gen AI to do stuff.
Yeah.
Like, I don't want it to just sit there next to me. I want it to do my tasks. And I think the thinking that I use is I got on a plane to fly here from Boston last night, and if I had gotten on that plane and it was just an autopilot and not a human pilot flying that plane- I probably would have turned around and walked off the plane and said, "I'll take the next plane. But if I got on that exact same plane, and there was a human pilot sitting there with all the 500 switches and controls that he has inside or she has inside that airplane.
Yeah
And they told me the autopilot was broken, I probably also would've turned around and gotten off of that plane and said, "No, thank you. I'll take the next one." Right?
Yeah.
I think that combination of an autopilot that does things automatically to make the human more effective, more efficient, and safer. That's what we're trying to build into the system with Gen AI.
Yeah, and where are you on that journey?
So we put out in 2023, a release that came out last quarter, the first set of capabilities that we have staged. So that is things that, like, allow a user to design a workflow simply by typing the name of the workflow. We'll actually lay out the workflow, lay out the data model, lay out the various user interfaces that are needed automatically. In fact, you can go to Pega.com and check it out. There's a, you can build.
Yeah,
It's a lot of things. We have use cases that allow people to talk with their data. So instead of writing a report to find out how work is getting done, I can literally ask a question, how many disputes did I process in Germany in September? We'll actually show you that break in real time. That's using Gen AI.
Yeah
To allow people to interface with the system in a much more natural and easy to understand way.
Yeah. And, Don, like, where, where are you, where does this run? Like, if you think about it, is that like in AWS? Like, is it in Azure, or is it you run it yourself? Like, how does it work?
So, I think where organizations are going to end up, it's going to be. Right? So we run Pega Cloud, and a lot of our Analytical AI, either on Azure or we've also added Google as an infrastructure option. Our clients want a little bit of flexibility-
Yeah
-in their structure choice. Our first release of Generative AI is using OpenAI models from Microsoft Azure. But we've architected it with the expectation that over time, you know, Google just announced Gemini today. Right? So there are new models that are dropping on that are, one, going to have different levels of efficacy for different use cases. So we want the ability to provide model in the right use case. We frankly also want the ability to apply the most cost-effective models for the right use case. You know, GPT-4 costs about 5x as much per token as GPT-3.5. When we need it, we want to be able to use it, but when we don't need it, we want to be able to use a less expensive model while we're prototyping work with open source models as well.
Yeah.
I think model flexibility is going to be a big piece of what the clients need to expect.
Yeah, yeah. And then I think you said at a New York event, where you kind of court a lot of clients, like, are there any examples or any early use cases that you could see that are kind of interesting?
Yeah. I sat down with the head of AI and analytics for one of our really large-
Yeah.
He said something to me that I thought was, which is, "Don't come to me with AI use.
Okay.
It's sick and tired of vendors coming up with AI use cases." What he wants is value. Show me that you understand what the problem is in my business. I have a problem that I need to put the right product in front of my customers right at the moment that they're ready. I have the problem that I need to make it so that my customers can get self-service for what might be a really sophisticated banking process with a lot of regulatory rules, right? Like, disputing a transaction with the bank. I want to make it so they can self-service themselves, themselves on their phone. That's my problem.
Yeah.
Now, show me how you can solve that problem, and if AI informs that and makes that better, and makes it easier for me to achieve that, makes me do it more effectively, or minimizes my regulatory risk, increases my efficiency, great. But don't show up at my door with an AI use case, because I know enough to know that AI by itself isn't the solution to every problem I have.
Yeah. Maybe this last question on that AI topic, because it is often in our circles, like, so how do you monetize it? Like, how does it fit in to the product?
We talked, we've talked about sort of three ways in which we're monetizing this. One is, we're seeing AI as a driver to cloud, the Pega Cloud-
Yeah.
... which is our software. The technology right now is moving so fast. Our clients are realizing they're not going to be able to actually do this on premise. Can't update them enough, they can't put them in security controls, they can't upscale them and downscale them enough. They need the cloud. So as our clients, we're using that as a driver to get the clients into our software. That drives a pretty significant ACV increase for us. The second is it actually drives more use cases. So the way we license Pega is primarily not user-driven, because if we're doing automation right, you actually should need fewer users in order to get things done.
Yeah.
We tend to license by what we call throughput, which is really the number of work, amount of work you put through the system. An address change is a case, or onboarding a new commercial banking client is a case, or when Verizon sells a new iPhone, that's a case, right? So we license by the number of cases.
Right.
So if I can accelerate the way in which you deploy new workflows in your business, that then leads to more cases coming into Pega, which should be more value to the client in terms of efficiency and also more security for us. And then the third piece, which you're gonna see more of from us in 2024, is actually monetized SKUs or actual products- Tied to specific Gen AI capabilities. So some of the optimization that I talked about, right? That would be an upcharge capability that a client would pay for, and that is additional product that we would sell to our client.
Yeah. You mentioned one aspect of AI that would help you nicely, which is like it needs to be more in a Pega Cloud rather than like the client cloud. Like, can you talk a little bit about what you've seen over the last few years, and then, like, you know, that extra incentive from AI to kind of move it to Pega Cloud? Like, where are we on that journey?
Yeah, I think the big thing that I've seen in cloud, and I think for a lot of clients, and frankly, I think we're gonna see the same thing with AI over maybe the next two to three years, is it is as much a comfort and risk issue as it is a technology issue. In other words, you know, we've seen the federal government go from, "No cloud, no cloud, no cloud, no cloud. No, it's got to be on cloud.
Yeah.
I've seen major global banks go from, "No cloud, no cloud, no cloud," to, "Oh, wait, now it's got to be on cloud. I think what happens is, over time, organizations realize cloud itself hasn't fundamentally changed.
Yeah.
What's changed is internal perceptions by the organization, that actually these cloud vendors can actually mitigate my risk and my security and my scalability a lot, frankly, in our own IT department. The benefit isn't there for me to try to run it myself.
Yeah.
So over the last couple of years, we've seen a real acceleration of clients moving to cloud, and clients at the scale that five, six years ago I would have said would never move to cloud are now moving. And that combined with our own advances in leveraging some of the latest cloud technology capability workloads at really significant scale have allowed us to dramatically move our margins associated with cloud.
Are you already at the point, you know, there's the carrot and the stick approach?
Yes.
Are we at it? It doesn't sound like we're at stick yet.
No, we're not at stick, and we generally try to stay away from stick.
Yeah.
Because at the end of the day, I don't want my clients associated with us, or my clients associated with carrot and doing things because there's value in doing it.
From a tech perspective, how much extra tech debt or tech headache do you get from customer on their cloud versus being clean on your own cloud?
The primary issue that we get is on our cloud, we get to upgrade.
Yeah .
We get to make sure that you're on the latest and greatest and taking advantage of the latest. When the clients deploy it themselves, it's up to them, and it's gotta get in line with their IT department.
Yeah.
So it's harder for our clients to adopt the solution that we're giving, which means it's harder for us to go in and convey to them the value they're getting.
Yeah. Yeah, yeah, yeah. Okay. Yeah. And then, if you think you're out of your market, and the movement market is like, it, like more and more coming together, as like the RPA-
Yeah
process mining, et cetera. Like, you guys are, you know, doing more process mining now as well. Like, how do you seeing that coming or coming together for you?
Well, I think-
For the market as well?
Yeah. I think we view these things as coming together for some time.
Yeah.
I've been in this long enough to remember when people thought that rules engines were a different technology that worked with automation engine. We got in a lot of trouble with Gartner because we were saying, like, you actually can automate workflows, have a really good rules engine, and there's no point in making a bunch of decisions if you can't actually do them.
Yeah.
Right? So we've long believed that a lot of these things. We think we're the first major workflow automation company to buy RPA for RPA endorsement. Because, again, we saw that, like, this is just another form of automation that's valuable if we can bring it together in one solution. Same thing with process mining. We really welcome the convergence of these, because at the end of the day, the value to the client is not in the technology itself, it's in: Did we improve this process to drive more efficiency? Did we make this user experience better? Could I create, automate a customer service process and connect it to a self-service channel, so you're deflecting calls out of the contact center?
Yeah.
If I have a richer set of technology that actually allows me to address that problem, I think I've got a better chance of solving the main problem.
And then, like, how do you see w hen you talk to customers and see them, like, how close is this market already coming together? Because like, you know, we had, you know, earlier, some of the guys that come from the other subsegments, and they're everyone is still doing fine, and nobody's really talking about, like, the, you know, the crossover yet in terms of competitive. Like, how do you, if have a conversation, how close are we to become, like, very much like one space versus everyone is still in this, like-
I think there's still, frankly, a little bit of confusion inside of the customer base.
Yeah.
You know, there's some degree to which both the vendors in the analyst market exploit that confusion, right? I've always found the best conversations happen with a customer when you're not actually aligned into technology categories. Because technology categories have a tendency to commoditize everything. I'd much rather drive a conversation around what are you trying to achieve? How the right technology together will allow you to do that. Let me prove that I can do that at the scale and the price, so that you get that economy of scale, and you don't end up with a bunch of silos and disconnected environments. When clients have problems like that, that's where they really turn to us.
Yeah. Okay. Last couple of questions were, like, again, like, on technology, like, the Pega Process Fabric-
Yeah.
Like, can you speak, like, what it does and why it's important?
Yeah. This came out of this desire to minimize silos in our clients. So there's lots of things in a big organization. that create silos, right? I sell multiple products. Great, I need an app for each product. That's a silo. I operate in the UK and in the EMEA, and I operate in Singapore. Well, every single one of those countries has data residency requirements. Silo, right? So there's just natural things in a big organization that create silos.
However, if I'm a relationship manager for Citibank, and I have a global client, where work for that client ends up in my Singapore system and in my UK system and in my Europe system, I have to physically stay there because data residency says it has to stay there. I don't want to have to go to three different systems to find out what's going on with my client. I want to have a holistic view that pulls that together, and we built Process Fabric as a way of accepting the fact that there just are going to be data silos in a big business.
Yeah.
But if I have a way to pull those silos together into a common view, that's my efficiency and the better experiences.
Okay, perfect. I think my time is up, Chris, but it was, it was really helpful to have this conversation, and
Yeah.
Thanks for joining us.
No problem.
Really nice to hear, like, the technology lens as well from Pega. Thank you.
Thank you.