Pegasystems Inc. (PEGA) Earnings Call Transcript & Summary

Good afternoon, and welcome to the last session of software today. My name is Blair Abernethy. I'm the software analyst and one of the software analysts here at Rosenblatt. Please help me welcome Pegasystems. We have Ken Stillwell. Ken's the CFO of company and COO; as well as Don Schuerman, CTO of the business. And, Peter, who is Peter Welburn, who is the IR Manager. And so we've got lots of firepower here with us at Pega today. I appreciate all of you guys spending the time.

So this -- you guys have participated in this event before. Our focus is around products and particularly around leveraging AI in your business. So we're going to start with a few questions, I think, for Don and maybe start with Don, just giving us a little bit of an overview of Pega from your position and sort of what you -- what your role is in the business?

Yes. So I'm Pega's CTO. I often describe that as Chief Translation Officer. I think a lot of what we do in a technology business is make sure that we can translate what is oftentimes a pretty sophisticated technology into real value that our clients can get. And that translation works both ways. Of course, we've got to make sure we're understanding our clients' needs and translating that back into our road map and our software strategies. And I think that's been especially true over the past 8 to 9 months as everyone in their CEO has become fixated on Gen AI and what it means for the business and how it's going to transform it's been interesting for us to step into this space, Pega is a low-code platform for AI-powered decisioning and workflow automation. So we've got decades of experience of using analytical AI models, business rules, automation capabilities like robotics and process orchestration and case management to drive mission-critical customer-facing processes for our clients at pretty significant scale. And as a business, we get used across large clients in financial services, in insurance, in health care, in telecommunications, in manufacturing, at the federal and large state level, in 5 big areas of their business in terms of how they engage their customers and really personalize that engagement to have very one-to-one relationships with the customer. How they drive onboarding processes and acquire new customers, new partners to think of things like KYC processes and banking, where there's a sort of regulated process that needs to get followed to onboard our new customer, customer servicing, large portions of -- or pretty much every customer service interaction in these organizations ends up in being a workflow that needs to be automated. So we drive a lot of the automation of those customer service workflows many customers have evolved that to being their customer service desktop and platform. Core operations or things like banking operations, health care and insurance claims, and then what I call resolving exceptions, fixing things when they go wrong. This could be payment exceptions, financial crimes, investigations, et cetera. So decisioning and workflow platform used across the customer life cycle by our clients.

Great. And Don, you've been with Pega for 20-plus years here, the companies had sort of a -- not new to AI at Pega. Maybe you can just talk about things have evolved there and sort of what you're really working on this year that's adding value for customers.

Yes. So just to go back a little bit, a lot of Pega's founding was based in the concept of expert systems, which are really rules-based systems. Some of the early generations of AI and using that to automate decisions in complicated business processes like payment investigations and payment disputes. We -- going back now about 10 or 15 years, we acquired in capability in what I would call the analytical AI space. So being able to use predictive models, machine learning models to predict and then use those predictions to drive next best action or the right conversations with customers in a highly personalized way. And that's going to be a significant part of our business. We have clients like Commonwealth Bank of Australia. People like to talk about this idea of private AI models. Well, Commonwealth Bank of Australia is running many thousands private and AI models that they've built using their own data but Pega's decisioning engine -- and those are machine learning models that are constantly getting feedback and improving the efficacy at which Commonwealth Bank of Australia and clients like them are able to engage with their customers. Have the right conversation, whether that's a retention conversation or a cross-sell, upsell, share of wallet conversation or maybe just a servicing conversation or even things like collections. We've taken that expertise and expanded it into other areas. So as I said, we do a lot of workflow automation. Well, workflow automation means we have a lot of structured data about how work has historically been done. So we've been able to use that predictive and analytical model to deploy process AI. So that we can predict, for example, when a workflow is going to miss an SLA and automatically drive an escalation based on that. And again, these are predictive analytical AI models that we're running at scale. In recent years, we've added in things like natural language processing, so we can automatically take in e-mails, convert that from unstructured and structured text. We've added voice AI so that we can take an automatically transcribe a customer service call. And then over the last 9 to 12 months, we've been looking at ways to integrate generative AI. So the next generation of these large language models that are focused less on analytical decision-making and more on actually generation of text and images. So summarizing texts or generating new text or generating new image. And as we've been doing that work in our upcoming release, which is going to be called Pega Infinity '23 and it will be GA in the coming weeks, we've got about 20, 25 new capabilities embedded in that release that explicitly use generative AI models to improve low-code developer productivity, the ability of marketers to personalize interactions via 1:1, the ability of customer service agents to better engage and support their customers, the ability of business leaders to get visibility into how work and processes are getting done inside their organizations. And so we're really excited about using generative AI to complement the existing analytical AI and automation capabilities that we have on the platform.

So that's great, Don. When you look at what you've done in the past leveraging AI and then now what you're looking at with this new release, with Gen AI and all the sort of features there. How are you monetizing that? How does -- if a customer has the Pega Infinity platform? Are they getting all of this? Do they pay extra for some of this? Because things like generative AI that comes at a cost on operational costs from your perspective as well, right?

Yes. So let me separate some of the analytical AI from the generative AI stuff just because I think they're important to understand both. We've historically had models for monetizing the analytical AI. So we generally do that ultimately based on the number of decisions that somebody that an organization has made. You make more decisions, you're theoretically getting more value from what the AI is bringing you and then therefore, we monetize based on that. And that's true for things like -- next best action. That's true for things like process AI, which I discussed. There's also been things that we've built like voice AI, which is an add-on license. So for our clients who have a customer service already, where we say and the ability to automatically transcribe a conversation is an add-on to that existing. As we look into the generative AI space, there -- we talked on a recent -- I think our CEO talked on our recent earnings calls sort of 3 ways that we see that being monetized. And I think the first 2 are more near term, which is, one, we're going to drive additional volume of cases into the Pega platform. So the way that we drive most of our contracts on Pega is what we call case-based. And you can think of that as like a usage-based pricing model at cases like a workflow. So a customer service request would be a case or a claim that you process or a new loan application. That's a case. So drive more cases on Pega you're driving more value and automation into your business, we're driving additional revenue or ACV based on that. And because the generative AI stuff, a lot of it will allow our clients to deploy more cases faster and easier than they ever have before, we anticipate that driving more case volumes for us. We're also generally -- Ken, go ahead.

I was just going to add just one add-on to what Don just said, Blair, which is I think -- so -- as clients use Pega more and more naturally as the case volume goes up, the economics get better in terms of relationship between Pega and the client. But the client also -- the more cases that the clients are pushing through Pega, the less likely humans on their teams need to interact. And so the benefit not just more volume, it's that their cost to handle those down materially because they're driving through an intelligent automation platform using AI and the actual human interaction, the percentage of cases that need touched by human hopefully goes down dramatically, right? That's the value proposition. So it's really the customer spend goes down materially our spend with -- our participation in that spend goes up proportionate to the value we're providing. So it's a win-win for us and our clients.

You're shifting labor, if you will, onto your automating more labor on to your platform. So they pay a little more to Pega, but their cost, their needs, or their capacity needs drop.

And the speed of the transaction goes up so that you're getting away from a client, a person waiting on the phone for 20 minutes and dealing with a human that may have -- and you're able to clear a lot of these actions using AI either by communicating with the client or quite frankly, communicating with nobody and just think straight through the processing as much as you can. So yes, that's the value proposition.

We're also -- the generative AI features that we're putting in Infinity '23 are only available on Pega Cloud. So our SaaS managed service offering. And we're seeing that drive a lot of interest in our clients about moving some of their existing applications migrating them on to Pega Cloud, which both drives ultimately lower TCO and faster response better for them, but that also drives additional cloud subscription and additional revenue for us. And then longer term, we also anticipate there being generative AI features that are add-ons to our existing licenses. So there will be features that in order to take advantage of them, clients may need to pay additional fees as well.

And remind us where your customer base is today vis-a-vis client managed cloud versus the Pega Cloud just to...

We're approaching 50-50 in terms of the percentage of our ACV right now, slightly less than 50% of our ACV is on Pega Cloud. But Pega Cloud is growing faster. And so in the near future, I would imagine based on just if you just look at the trend, the Pega Cloud will overtake Client Cloud. And that's all happened, Blair, in the last 5 years. So it's been a pretty fast chip.

That's great. And I think, Don, if you could just maybe describe a couple of things for us here to help us understand. So you have announced recently Pega Gen AI as a product or Pega Process AI. Can you just briefly describe what those are?

Yes. So I'm going to actually start with a take on Process AI. So Pega Process AI is the analytical side of AI. So if we think of AI as kind of being left brain, right brain. So analytical AI left brain AI, makes decisions, right? Next best action decisions, process optimization decisions, et cetera. Most clients want that to be built on their own proprietary private first-party data. They don't want prebuilt public models. They want that to run on their own data because ultimately, they view those decisions as differentiating to their business. Differentiating in how they engage with their clients, differentiating and how they find optimizations and how they run their workflows, et cetera. So process AI basically allows clients to take the data they already have in Pega. From business processes, historical cases that they've run through and find based on that data predictions for how future processes might be. So imagine being able to predict at the outset of a payment dispute that regardless of the amount of effort you're going to put into this, you're going to write this off. Well, if I can predict that with 90% certainty the day you get it, and it's going to cost you $50 to write it off versus $100 to actually process it, why don't you just write it off, right? And that's a net saving. Or if I can predict if there's an 80% likelihood that this is going to miss a regulatory deadline, which results in you getting charged to fee, and I can predict that a week before that happens. Let they automatically escalate back so you avoid actually having to pay that regulatory fine or missing an SLA with a customer that's going to degrade the customer relationship. So that's the use case for process AI, and that's an add-on to our existing case-based license for clients. Pega Gen AI is a series of capabilities that are coming out. And say by the way, process AI is generally available. It's been generally available for over a year. So clients have been using it. Pega Gen AI is a set of capabilities that are coming in Pega Infinity '23, which will be GA in September. And it's about 20-plus Gen AI-powered -- I would think of them as accelerators or boosters. And based on feedback with our clients is how they're thinking about using this Gen AI stuff. One of the things that they're leaning into are human-in-the-loop use cases. I don't think many of our clients are ready to set Gen AI loose to directly interact with our customers or to do things without a human in the loop. So we've really focused on these human-in-the-loop use cases. So for example, being able to build a business process or workflow simply by typing the name of the workflow into Pega. So I can just tell Pega to build the student loan application workflow, and it will come back and say, great, that should generally have these 5 high-level stages, here are the 5 steps that would happen under each stage. This is what we think the data model should look like. Do you have any changes to that? Or do you want me to build an app that does that? And the great thing is it accelerates the development process because that might have been like 2 or 3 weeks of requirements gathering and debating, and horse trading among team members. And now we take you from a blank page to something that you can really work with. But it still gives the low-code developer of workflow analysts all the power they need to change it, override a suggestion, add their own steps, remove steps, change the data model, add the data model, right? So I'm accelerating the process and also lowering the bar of entry. So that somebody -- you don't need a huge amount of Pega experience now to get started with building your workplace, we the type the name of the workflow, and we'll start it for you. Other examples you build a workflow and you're going to do this at an enterprise scale. You need to test it. But, that means you need test data. So generative AI is really good at generating a bunch of test data, and that's a no value-add ask that I now can take off the developer so they can focus on adding business values to the application. Another use case for generative AI that we've got in Infinity '23 is in this next best action space. So one of the things that happens when customers shift their marketing from being kind of spray and pray traditional campaign spam to very one-to-one personalized is the number of treatments that they actually need increases because I'm no longer just giving everybody the same ad. I want to be able to say, like our millennial. I'm going to talk to you differently than I talk to a family that's just had their first child. They're somebody who's about to retire. And so that means that a bottleneck often becomes the ability of marketing to generate all these treatments. Well, Gen AI is really good at saying, "We've got an offer for a new loan, write me a treatment for a millennial audience", right? The treatment for a couple of the new baby, right? Mean treatment for somebody who's [indiscernible]. And then bring those back to the marketers so they can tweak them, override them, make changes to them. But again, I've accelerated the process of rolling out those new treatments, so I can have more personalization at scale faster than ever before. And then we've got also a series of use cases that fall more in the customer service space. So a lot of -- I don't think any of the organization is ready to put, for example, ChatGPT directly as a chatbot out talking to their customers or any kind of generative AI. It's not predictable enough. If you ask it the same question 5 times, you'll get 7 different answers, right? There are really good natural language processing-based chatbots that organization is very comfortable with converting with our customers. The challenge is you've got to train them. You've got to tell them that there are 100 different ways to say what's my balance in Dutch. Well, the neat thing is, I can ask generative AI to come up with a list of 100 different ways to say, what's my balance in Dutch. And then I can feed those 100 different ways into my NLP chatbot and now it knows that. So I've accelerated how fast I can train these bots to make them more responsive. So those are the kinds of use cases where we're using generative AI to accelerate how you get automation and real-time decisioning in front of our clients' customers and in front of their employees faster.

Blair, just to add to that, one of the -- if you look at the way that the chatbots operate today, which could be mistaken for being AI-based because they do seem to have a conversation with you. But really, it's very structured. It's like looking for certain responses. And if you say something, it will say, did you mean to ask A, B or C? And you click, it's really a decision tree model, right, in terms of how it goes through that. And we -- and that is helpful but that isn't Gen AI, right? And that's like a I think clients are nervous about just saying, "I'm going to put it out in the wild, I just left the act. Right now, it's kind of like almost -- it's really not AI, but it feels like AI. But the first step was, I'm going to build a decision tree and it's basically going to be like one of those novels where you like pick and it says, go to Page 88. But I think that when we get to the real world the use case of Gen AI like we know it to be, like where it will actually sift through and give you a conversational answer you really need to make sure the data link that you're using, you got to do bit of a testing. So I don't think we're a little ways away from people taking that kind of risk on really deep relationship clients because you don't know if you're going to offend them, be in the wrong language, make the wrong recommendation, tell them, "Oh, you'd be better off to buy the competitor's product", right? I mean, you don't know where that's going to go. So I do think that that's the one I think people are really optimistic on. But I would say that -- that is where clients are a little nervous, [ the right amount ].

One of the used cases that we've also got an Infinity '23 that I do think it's really interesting is, more and more of our clients are seeing their customer service volume switch from voice to chat because everybody wants to chat. And it turns out a lot of the customer service agents that got really good at voice that doesn't naturally translate to being really good at interacting with chat. So they need to be trained and they need to practice. Well, it turns out that ChatGPT does a really good job of simulating a customer in a chat. So we're going to have in customer service with Infinity '23, the ability to use generative AI to act like a customer with a specific personality, maybe they're angry because they've been waiting on the phone or maybe they're really exciting because they really like this new product. And you can have that customer with that personality simulate a chat interaction with your agents, so you can train your agents faster and get them more comfortable with handling the increasing chat volume that many of our clients are seeing. And that's a place where I feel like, again, it's helping the human better. And I think our clients are more comfortable with that kind of use case because it doesn't have the risk of like going directly to their customers and saying something they didn't predict what's going to happen.

Yes. Have you had -- what have you done so far in terms of beta testing of some of these new capabilities?

So we've actually had, out in the wild for a while, what we call the self-study body, which is basically we took a lot of our knowledge content in our documentation. And we used -- without getting too technical. We used -- there's an architecture for using GEN AI called embedding. And based on what you do is you use -- you do prompt engineering and you take an existing set of knowledge and you use that knowledge to write the prompt to AI to the Gen AI model, where you basically tell it -- I want you to answer this question, but I only want you to answer it based on this knowledge. And if you can't find the answer in this, just say I don't know, right? And so it's a really good way of using Gen AI, but restricting it to a known knowledge base. So we've had that out in the wild with our own documentation and our community and support. And we've seen really great responses from our client developers, from developers and our partners because it allows them to quickly go and ask questions and get answers. And summarize what might be like 10 pages of documentation and turn it into like a 5-step checklist, right? And so that's been a really kind of nice forward-looking indicator for us that if you do this in a way where, again, you're empowering the human and you're keeping them in the loop. You've got a degree of traceability. The other nice thing about this is the buddy can actually give you its references. It can actually tell you like, "by the way, I got this answer from this page or a source website". So you can check it, right? That we found that when you have it with that kind of credibility and that kind of safeguard built in, it's been really embraced. So we're excited about that.

Blair, our region -- I'm looking at the sale. I'm looking at the buddy one that's [ Downsell ]. I'll read you like just a question that was asked this morning. Write me a short summary of what Pega Customer Service can provide, make sure that you differentiate it from our leading competitors without naming these competitors explicitly in the node. Like that's a -- put and that gave about a 10 line response that you could use like that's the kind of stuff like -- if you think about the -- where it's not pulling from the public Internet is pulling from our content. So it's going to -- and it's -- that's a powerful thing to use in a service channel.

So you're helping to pretrain before you release it to your customers so that it's more effective for the guys that are actually going to be building the processes, right?

That exactly right.

Ken, let me ask you this question and this technology is fascinating and powerful. What are you -- what is Pega doing in terms of looking at using some of the stuff internally on your turning it back on to Pega.

Well, what I just showed you is actually something we're using internally to train and enable our sales team. We're also -- we got to be careful that we don't push too many of these initiatives, right, because then it becomes like a little bit of -- you get -- you're spreading your value across it. So we're focused on enabling sales as an example, right, things like writing script, inviting people to PegaWorld, things -- those were some of the test cases. We've also we're also pushing this into our internal portal, which is one of the initiatives we have, which is people employees when they come to the portal, and they want to -- normally, they would go to a drop-down and they'd say, let me go into people hub because I want to change my 401(k) election, so there's a click and then another click and then a click to a third party as opposed to saying, "Hey, I'd like to change my 401(k) election", being able to interact saying, "Oh, here's where you want to go." Naturally, the next step would be, what would you like to do, and we'll go ahead and execute that change for you. But I think the first starting point is just serving up the information in a more kind of question-and-answer conversational environment. And that's a very big kind of experience change for us internally. And then I think the last, another frontier is really around how we -- Don talked about quick starts and how you use general AI to get frameworks that you can actually use it not just to create an app, you can also use it to modify or improve an app, right? Because if you can -- so that's a -- and we use Pega for lots of applications. There's probably 50 use cases inside Pega that we use Pega to actually execute those internal processes. And we have teams that do application improvement around those. And actually, that will speed up the throughput and the efficiency of that as well. So those are some examples where we're using it internally.

That's great. That's great. Don, if I can take it back to you for a moment. One of the things that you guys launched about a year ago was kind a Pega Launchpad. Maybe you could talk about that a little bit. So where you're at with that initiative and what it could do for your customers?

Yes. So the idea with Launchpad was we have built a platform that was really, really good at workflow automation and AI-powered decisioning. But it was really targeted at pretty big substantial often global organizations that has internal workflows they needed to automate. And we were seeing increasing demand from a lot of our partners. They wanted to actually use a lot of the core capabilities in Pega's platform around workflow automation and being able to build U.S. experiences across channels, et cetera. To deliver their own SaaS solutions to market. To come with like a pre-built workflow-based product that they could again go sell out to their customer base. And there was a need where some of the needs for that market are actually slightly different than some of the needs for the market that we have. Our Pega Cloud is largely single tenant and our clients like that because it means we can like give them their own private data and give them security and VPN into some of their systems when necessary. Our partners who wanted to build solutions on Pega, they actually want multi-tenancy because they want to be able to operate at scale and have flexibility to scale this up and down with the kind of cost margins. So Launchpad was designed to provide a platform specifically for SaaS providers who want to then build and take their own workflow-based applications to market. And so we've been working over the past year with a number of early adopters as they go through the initial development phase of those applications. So that we're there -- so that they're getting close the point that they're able to then launch them how it is a public as sort of initial to their initial early clients.

What's the revenue model for Pega from Launchpad?

The revenue model for Pega is essentially a revenue share model which is we will work with the intermediary, which is the ISV as we might refer to them, where they will be the ultimate vendor to those clients we will be the platform that they operate on, and we will help them think about the commercial models with those use cases to optimize the model for them, and we would then get a share of that revenue based on that kind of flow through.

Okay. Great. And so is it -- it's GA now? Is it -- or is it still in data?

It is a -- we have a number of early adopters. It is available but not to anybody just to the specific -- we've got essentially less than 10 that we're working with that we believe are -- that are excited about it that we think have the requisite Pega skills to be able to support it going after. So I think 2023 and into '24 is, I would more say it's a selective release as opposed to the term GA.

Then eventually will it have the -- some of the AI capabilities on and Gen AI capabilities that we're talking about earlier.

It's Pega Cloud. So it would have -- it would have accessibility to anything that Pega Cloud does, yes.

So only limited by the imagination of the ISVs.

Yes, that's right.

That's great. Maybe shifting gears a little bit again. One of the things, Don, you and I talked about earlier this afternoon was this concept that was described as the age of the autonomous enterprise. Yes, which is a longer-term vision, but it looks like the pieces that are being put in place to make that happen. Maybe you could just describe that a little bit for us and help people understand what you mean by the autonomous enterprise.

Yes. So I defined it -- let me define it and then I'll talk a little bit about how we see this evolution happening. So I define the autonomous enterprise as being able to use a combination of AI intelligence and automation to drive a business that is continuously self-optimizing. So imagine a business process that is continuously finding bottlenecks and breakpoints and actually fixing itself. Or the very least, recommending out to the business, here's how and where you should invest in fixing and improving the process. Or imagine a customer interaction that is learning in real time from previous customers and actually getting smarter on what products it offers. In fact, in some of this -- we have clients who are doing this today. So many of our clients who use Pega for next best action, use a product called Customer Decision Hub. They have models out there that are continuously learning and predicting, for example, what customers are likely to churn and making a recommendation to save them and learning from what works and getting better at it over time. All with guidance from their data analytics teams and marketing teams and customer experience teams. So the autonomous enterprise is about carrying that vision forward. And over the past number of years, we've been putting together the kind of pieces of this, like we acquired process mining last year because that's important because that gets you the data about where bottlenecks are, not just in Pega processes but in other processes that are maybe outside of Pega. The work that we've done with process AI to drive continuous optimization of the business process and predict where processes are going. What's really interesting with generative AI is it now begins to stitch all those pieces together. So I talked at PegaWorld in my keynote about this idea of an autopilot, right? So imagine a business leader being able to say, "I want you to optimize my claims process. And I want you to do it in such a way that reduces processing time, but maintains my customer Net Promoter Score." And now I can go out and look, the autopilot can go, and look and say, okay, I looked at process mining data, I found bottleneck I think I can fix that bottleneck by inserting a process prediction that skips over when it's not necessary. Do you want me to do that? I found a screen where you're constantly having to go back because the user is in putting incorrect data. I think you should go have a UI designer look at that screen and fix it so you don't have to keep doing this rework. You want me to sign a task to a UX designer to go fix that screen in? Great. I'll go do that. That's now a process that's continuously optimizing. And where I think of like where the next value sort of stage escalation of value from generative AI comes, it's that. It's pushing businesses into the self-optimizing space where not only are they getting automation value right now, but they're continuously finding ways to improve efficiency into their processes, or improve the efficacy of how they engage with our customers to do things like drive revenue or maintain their existing revenue streams.

It's interesting. I spent a lot of time at your event this summer looking at the process mining capabilities that you acquired. And maybe you could just describe what that is and -- because it seems to be an area that is a real value creator, I think, for customers.

Yes. So we acquired a company that provides basically ability to look at historical process data. And from the historical process data, find bottlenecks, to find places where the process is getting stuck or taking longer than it should. Find places where rework is happening, find places where the process is like diverging from the ideal path to the happy path. And one of the reasons we really like the technology, the company that we acquired was, they already had built in some -- not just the ability to diagram and draw the process, but be able to do like root cause analysis and tell you like, this is broken and here's how to fix it. And so we spent -- since the acquisition, we've been making sure that's really tightly integrated into the Pega platform so that -- if you've got an existing process running in Pega, all you need to do is turn process mining on, and we'll start finding new opportunities for you to make that process even better. Additionally, it can import process from other systems like SAP. So if you want to take in your SAP data, you can also do that. But the important thing here is, I think this is distinct from how other process mining has come at it. We come at it from the endpoint of ultimately, we want to execute this business process so we can automate it for you. We don't just want to mine it and find like opportunities. We're fundamentally in execution on an automation engine. Process mining is now just an input into that engine so that it can find continuous optimizations and waiting to get better. And I think the intersection of now having the process mining data, plus our core capability in being able to drive automation and AI decisioning into the process. That just is a huge value opener for our clients.

So you've got a lot of innovation happening this year and this past 12 months at the business. How are you getting that out to customers? How are you educating them and getting them to really start to pick up and utilize the power that you're putting into the platform.

Yes. So one of the things, I think Peter has talked about is really making sure, one, that we've got focus on having very deep and intimate relationships with a select set of customers. So one of the things that we're really focused on is the large-scale organizations that really have the opportunity to get great value from this where we're already a proven trusted partner. That gives us the opportunity to make sure that we are in the door walking the halls, advising the centers of excellence and lead teams so that they're continuously driving and taking advantage of this. We're also continuously making sure -- like one of the reasons why we're looking at things like generative AI plugging into our documentation and plugging into our enablement and training, so that we continuously provide ways for our customers to self-serve and self-enable on this stuff. So it's very easy for them to go and find out what are the latest and greatest capabilities. How could I use this with my business process? And more and more, I think, as Alan kind of hinted at his PegaWorld keynote, he talked about having Pega expertise at your fingertips. I think the neat thing is by integrating some of the stuff that we've done now with generative AI and our content and knowledge base directly into our development environment, we could have to take it exactly to where our customers are. So they don't even need to go looking for it. We're just kind of prompting them right in the design experience, like, hey, by the way, you're taking new capability or you can use this, you could do that. And I think that, combined with the very deep and kind of personal relationships we build at a client-by-client level really allow us to help ensure both that our clients are getting value. And frankly, we're getting the kind of feedback we need from them to continuously improve and make sure that what we're delivering is the right thing for their needs.

It seems like these -- adding this -- is this additional intelligence on top of your platform or into your platform in various areas should really make the product stickier as well over time. We should have not?

Yes. I mean I think I'll let Ken and Peter speak to our retention rates. The thing that I would say is we tend to do pretty mission-critical processes for our clients. And those mission-critical processes tend to be heavily integrated into their existing systems, into the existing channels and front ends. And the combination of that plus the fact that we've now got this data that we can help our clients in mind for increased optimization and increased efficiency, I think, helps ensure that we are pretty sticky with our clients.

I think that we have historically had very high retention rates, Blair, but I think the difference with what this does is, I think this -- we become a potentially a very critical partner to our clients initiatives around how they drive efficiency in their organization to leverage AI. Resources are very scarce both in dollars and in people. And it's very hard for companies to -- in today's world to just say, "Well, I'll just hire another 300 CSRs. I'll just hire another 500 [ assessors ]. Like those days don't really exist. I mean those resources are scarce. And just with the way the world has changed, they're not even co-located in many cases in the same look, you didn't get the benefit of having everybody in the same building, people are working remote, they're all over time zones, kind of the [indiscernible]. So you really are very dependent on the next person you need to hire, the next -- and that's a very fragile business model when you're trying to support clients in digital channels, like Don mentioned chat earlier. I think that's the pivot that our clients are making -- is really making this autonomous kind of enterprise like kind of analogy is be real, right, where they can actually say, "You can talk to us any time you want from anywhere around anything, and you're not going to have to wait to get in the queue to get something on the one with you."

And I think that's especially true for some of like the very specialized roles that clients need. Like they are never going to be able to hire enough data scientists to like manually build the AI models they need on their own. So the fact that we can come up with like Customer Decision Hub and process AI with an engine that can actually look at the existing first-party data they have and automatically create a self-learning model that can then optimize the customer engagement or find efficiencies and predictions in the business process means that even without having a team of dedicated data scientists looking at this problem, we're helping them surface efficiencies and get benefits from the AI technology. And we're doing it in a way that's proven to be transparent and explainable and protects their data and test for ethical bias and all the stuff that clients know they need to do. We've got a decade of experience doing that really well. And I think especially when clients look at like, how do I turn all this AI from a science project into value? We've got a history of doing that, that doesn't require them to find 100 PhDs to make it happen.

Yes, which are expensive and scarce.

Yes.

All right. Well, listen, I think we're up against our time here, but this has been a fascinating discussion. Don, Ken and Peter, thank you very much for taking the time out to give us a little more insight into what Pega is doing with this technology and looking forward to the full release here of the next Infinity platform.

Thank you.