Pegasystems Inc (PEGA) September 8, 2026 Conference Transcript

Well, thanks, everybody, for joining us this afternoon. I'm Steve Vendors, part of the software research team here at Citi. I want to welcome everyone to day one of the Citi Global TMT Conference. With us for this session, we have Ken Stilwell from Pega.

Ken, I want to thank you so much for being here. Thanks, Steve. Appreciate the invite again.

Yeah, of course. Maybe just to start off, you know, I'm sure people know who Pega is, but maybe we can maybe introduce Pega investors who maybe are a little bit newer to the story, and maybe what are some of the key problems that PEGA is helping customers solve today?

Sure. So PEGA has been helping enterprise clients for, gosh, four or five decades. And the types of solutions or some examples of the things that we do to help our clients, it typically referred to as enterprise workflow and specifically workflow where it's deterministic, which is things that either have a certain process that needs to be regulated or there's a requirement for a certain way that something gets done, or internally clients may put those types of controls on their own internal processes. So, for example, things around ERP as one example, that might not be legally regulated, but you may have certain internal controls or processes that you need to have repeatable, that need to be very predictable, that can be proven after the fact that you followed a certain series of steps and stages to be able to execute that work. And we've always had AI. Pega was – we actually purchased a company in 2010 called Cordiant, which did statistical AI. We integrated that in with our solutions as a customer decision hub of sorts, with many people that know Pega will have heard that term before, where we helped drive certain statistical probabilistic-based decisions or outcomes. With generative AI, which we've introduced into the platform a few years ago, we're now leveraging gen AI both in the design and the runtime capabilities, design being heavily when you're conceptualizing and modernizing a specific application and runtime where you want to really try to take unnecessary human steps out of that workflow or just otherwise make it be more automated, whether that be kind of the interaction with the customer or in just terms of the back office transaction flow. Okay.

All right. No, that's helpful. I do want to talk about AI and what you're doing there. I guess before we get into that, I do want to talk a little bit about just the demand environment and the performance that we've seen so far this year. I think for the quarter, we were at 8% constant currency growth on ECV. You were originally talking about about 15% growth for the year. How does that maybe performance that you've seen so far this year change your outlook, or how are you thinking about the ACV environment in the back part of the year?

Sure. So one thing to start with is we only guide, for those that are not aware of this, we guide at the beginning of the year, and we do not re-guide. We don't guide quarterly. We guide for a full year, we don't re-guide. In that environment, what we tried to do is to help provide information that would help connect some of the views that we had around where the full year looked like it would land. Many of our sell side analysts, including yourself, took that information and then kind of reprojected what they thought the year would look like. And I think you guys have done a good job kind of collating around where that was going to be. So one, I think the reason why we're not going to get to 15% or I would say it's unlikely we're going to get to 15% is we already had a lot of business in the back half of the year. And for us, we don't have a lot of pipeline build and close within a 30, 60 or 90 day period. Our campaigns tend to progress over a quarter or two or three. So for us to close something in, say, Q3, it likely would have had to have been in the pipeline when the quarter started. So we just have like a runway opportunity. The second point is that the ramp that we had in the back half of the year over what last year was a pretty significant ramp already in terms of the growth in net new ACV year over year. We were kind of already targeting something like in the 30s in terms of the percentage increase year over year. So we had a pretty aggressive ramp. We don't have a very short pipe building to conversion process. And what we saw in the first half of the year, I think probably we didn't do as good of a job in the back half of 2025 and maybe even into Q1 to try to think about leveling the year more, pulling things forward that we could, building back up pipeline bookings, et cetera. So for the full year, I think the street has a somewhere around 10% growth or something. And, you know, if I had heard what we said on our call, I probably would have landed somewhere around that number.

Okay. All right. That's good to hear. Maybe we can talk a little bit about the levers that you feel like you can pull into the back part of the year. What do you feel like you can do differently or what changes have you maybe kind of put in place to improve the execution and improve the sales performance and the back part?

So a pretty big shift for us. we intended to embark on at the beginning of the year at our sales kickoff. And just to be honest with you, I don't think we fully internalized or pushed hard enough on this change. And the change was historically, Peg has been more of an account manager type selling motion, which is we have existing organizations. Those organizations spend with us. We look at expansion opportunities. We don't have a lot of new logo chasing. We don't have a lot of growth even outside some of the buying centers that we've typically done business with. That has not been something that was, I would say, reflexive at Pega. When we did our sales kickoff in January, we instrumented a number of activity measures that we wanted to measure, things like new contacts, new meetings, more kind of think about them as more hunter or outbound type activities. And one could probably argue that we should have probably always had some level of that motion, but we really wanted to intentionally do that in 2026. A few reasons. One, we were starting to see a little bit more interest in new logos. Two, we knew with AI that with AI, everybody is telling a story with our clients around AI and how they're going to solve all their problems. We had to make sure that we We were kind of instinctively going out and getting that message in front of clients. And it turned out that in Q2, there was a lot of confusion with clients around AI. What's the value prop? Every single week, somebody would come out with some new industry that AI was going to put out of business, right? So there was still a lot of talk, a lot of hype, a lot of anxiety, a lot of confusion. And at that exact time, that's when we needed to execute on our hunter mentality. And I just don't think we did a good job of executing that. So we really hit hard toward the end of June, which quite frankly was probably too late to actually make a difference in Q2, to make sure that, listen, even if we're talking to the same buyers that we know, even if we're talking to our contacts that we may have known for 25 years, they're getting attacked by all different vendors, whether they're AI natives, whether they're competitors or service providers, talking about the promise of what they can do for our clients. And we need to make sure that we're with them and helping educate them and arming them to understand how we can help them and how our unique value proposition. So I think there's – I would call that in the bucket of execution, which is managing activity, managing contacts. And I think we've now – we've instrumented that. We've actually now put that into our sales forecast calls because we think this is not something that's going to be a one or two quarter. We believe this is going to be critical to make sure that you're constantly building like the motion of getting in front of clients and talking about the value problem. Now, interestingly enough, we have Blueprint that we released a couple years ago that's been a great tool for actually enabling that conversation with our clients. So the key is Blueprint to activity, getting in front of buyers, whether those be buyers we know, buyers that we don't, new logos to really drive kind of the awareness of how Pega can help. And then the next naturally tracking pipeline build, pipeline progression. But the most important first point is really just getting, you know, not taking anything for granted around our clients' digital transformation journeys, even if we have very, very deep relationships with them. Okay.

And I guess since you've started putting this emphasis in place and trying to change the mentality, what have you seen so far from the sales force? Like, what are the metrics you're tracking there? How are those kind of trending? And just, you know, how is that maybe looking through Q3 so far?

It's interesting. We've had, just as a small start to answer that question, we've seen some people in our sales team that have said, obviously, thank you for reinforcing that. Of course we do. we've had others that have kind of paused and said you know it's are you sure you want me doing that you know like i have one big client i have two buyers and so what we've realized is that that's that was good feedback to understand because you know we're not saying that every single of our people on our team is are going to do hunting and not account there is a balance of how we distribute the team so i think we learned a lot what what i took from that though is the people heard. They were processing it. We had those. So that was great. Our sales managers, I think also keeping in mind that this is not just about an account executive. This is about the whole management chain, right? If you have an account exec that's really good at hunting, but a management chain that isn't, that won't work. If you have a management chain that's hunters, but an account exec that isn't, that won't work. You have to really have, everybody has to be kind of moving to the same beat. And so I think from that standpoint, like that was a really great like anchoring and awareness to see how much our teams internalize the importance of that all the way up to our, you know, our chief revenue officer. The second thing I've noticed is that the amount of engagement and insight that we have early in the sales campaign, you can see the direct connection where that activity is higher. So I think the, and that's exactly what we wanted, right? We didn't expect this to be like, you know, a silver bullet that when you did something, pipeline would just immediately grow. This is a part of a process of really being deeply engaged with buyers and influencers. And PEGA is the kind of solution that does have, you do at times have a buying by committee dynamic of it, right? You have a business owner, you have someone in the CIO's office, you have other standard, you have a system integrator, you may have executive sponsors. There's a lot of influencers in that. So I think the thing we've just seen as the is two things one is the internalizing why we're trying to do it i think was um you know was uh is a big change but i think has been um internalized by our field teams and the second the second is um is just the uh you know the activity of actually getting those at bats right and by the way we just released infinity 26 uh i think it was ga about 30 days ago and we what better thing to do to use that as a way to get in front of our clients and get them exposed to the to the ai you know to ai native solution that we just released i do want to touch on touch on that before we go into that though um maybe just last question tying a bow on

you know the sales execution side of it it did seem like there had been a big focus on net new logos and trying to drive uh drive that and maybe a little bit of shift away from uh the existing installed base just you know with the pipeline build and how long it takes for some of those deals to progress and get through the finish line. Just how does that maybe change how you think about the mix of existing versus net new through the rest of the year and maybe into next year and just the pipeline dynamics between those two different customer bases?

So it's a great question about where do we think the opportunity is. And I think that I would maybe draw a bigger circle around this and say it's not just new logos, it's all new workflows. It's anything that's new to what we've already done. That could be a new workflow within an existing buyer that you might have had a 20-year relationship with. It could also mean a buyer that's two doors down in a different business unit at that same client. And then, of course, net new logos, someone that doesn't know Peg at all. So I think that I would say all of those are important to get that outbound motion. We would expect that what would come from that over time is that new logos would start to creep up a little in terms of the total amount of the impact on growth, although that may be small enough and happen over a period of time that it may not be that noticeable in terms of the ratios. So we're going to be very dependent for our growth to accelerate and continue to accelerate. We're going to have to sell to new use cases that we have not sold to, and that involves in a company like Citi, and it also involves to a financial institution that we've never done business with.

I do want to make this interactive. So if there's questions in the room, we'll make sure to get to those. I do want to ask a little bit about the AI strategy first, and then we'll make sure to get to those. So just on, you know, the new Infinity release, and, you know, I think that it includes Infinity Studio, what's different that that enables for you in, you know, both from a customer standpoint, like what they're able to do that they weren't able to do before, and maybe how does that kind of augment how you think about the ability to win or right to win within some of these accounts from that?

So Blueprint, I'll talk about Blueprint and then Infinity Studio. So Blueprint, the purpose of Blueprint was to get into the design and the selling activity where we could help our clients visualize and ideate around ways that they could visualize how to solve a kind of a reimagine or a transformation of a legacy application. If we didn't have Blueprint, the way that was done was through a whiteboarding exercise, typically. So it involved a lot of what I would say custom and unique interactions that were not leading to necessarily as fast as we would like them to. So Blueprint was around a very structured, simple set of dropdowns and a few fields to be able to sketch out what that design experience, that upfront design experience. But without Infinity Studio, without Infinity 26, that left clients with this concept. Here's a concept of something. I could kind of see what it looks like. How do I make that real? It wasn't as easy for them to make it real. they would go into a development experience that was like the development experience over the last 25 years. So that's where we had to really evolve the development experience, which is Infinity Studio. So think of Blueprint as I'm going to ideate. I'm going to build kind of almost the wireframe, so to speak, of what this is going to look like. I'm going to load that Blueprint into Infinity 26. And then in Infinity 26, I'm going to have an AI-assisted actually development where I can say, okay, I'm trying to build this customer service app. Here's the vertical that I'm in. Here's what I've done so far. Tell me what I'm missing. And it'll go through a series of things. Have you thought about your integrations? Here are some likely integrations. Do you need an MCP connection to be able to have people use AI tools to be able to interact with the design environment for your application? Like you could go through and it would help you. It would really help you not get lost, right? It would help you not be lost in that journey. And that's really critical because our clients don't want to have to be super users to be able to go through that design experience. And historically, companies like Pega and Pega as well really had levels of higher certification that you were required to get to really be productive in the development environment. And that's – so we wanted to change those two dimensions. One is at design time, kind of in that upfront blueprint, and also at design time when you're finishing or building out MLP, one of the minimum level of the product and evolving it, innovating it, we wanted to make sure that that experience didn't require you to have to have a PhD in Pega. Okay.

It seems like what you've been trying to do is accelerate the development lifecycle and make it faster to get Pega to production within clients. I guess, where kind of are we on that journey? you know and when i think of like pega historically it was very kind of maybe cumbersome to to maybe actually get implemented and took you know a year-long process just how much faster is it now and what more can you do to you know make it even easier to get those opportunities over the finish line for for customers and faster means not only time to value but also the cost of value just to Just to add on to what you're saying.

So I want to give you an example. We ran through an A-B test, so to speak, of like, what would it be like if we did things the way that we've historically done them? And what would we do if we actually took our engineering teams, keep in mind these are engineering teams, and basically said, let's build a demo app. Let's how long it takes from Blueprint to something that was production ready. Now, admittedly, I'm not saying this is production ready like we had all the integrations tied out, but it was far enough along in terms of that. And we did that with our engineering team and our own PEGA teams internally to be able to, and using the traditional way that we would have done it, it would have taken almost 2,000 hours to build an application. Our engineering teams were able to build that application out in 45 hours. so that that that is an example of how much faster it can be now that's our engineering teams we need to make that so that our to that our clients so the people that are not but what you can see is the innate capability that you actually have and how fast that can be so this is a little bit of a technology and change management you know challenge it's not just about technology but But I do think what that highlights is, like, you can get to a production-ready application in a time. Time is actually the amount of time the technology should not be the hurdle. It's going to be more the adoption and the change, and that's what we need to focus on.

That's an interesting point because, you know, there's all the talk about, you know, forward deployed engineers, all the talk about, you know, leveraging resources from one area, moving them to another. Just how do you think about, I guess, using the services as kind of like a go-to-market function versus, you know, maybe utilizing that for cost savings or reassigning those folks in other kind of areas?

So I think that the forward deployed engineer type model, whatever you want to call it, I know that we don't refer to it exactly that way, but that model is, I would characterize it as a necessary evil right now for a lot of companies, right? Because things are not inherently as easy as you'd like it to be. I mean, for those of you that would go into Clot or OpenAI and, you know, there's things that are easier and there's things that like you have to kind of just like grind your way through trying to figure out you probably had done it 10 different ways that are wrong. And then finally you figure out like, oh, I realize that's an enablement challenge. So I think with a lot of the FTE models, it's like let's just take all that away from a client and let's just do it for you. Okay, that is certainly helpful to demonstrate value and get there faster. That is not a scalable model because those FTEs need to stay on with those clients ongoing. And that's not like you can't monetize that. A client like Citi is not going to want to continue to pay for that. And there's not enough margin to build in to be able to handle that. So I think that is a helpful but insufficient way to solve this. You've got to use some of those professional services and sales engineers to get the momentum started, and you've got to have a product that intuitively people understand how to actually build and evolve their workflows. And I think that whole AI assistant, that agent capability that will sit natively inside something like Infinity 26 for Pega is a critical component of how you do that because it has the whole knowledge base that no other agent has that knowledge base because it knows natively everything that pega does where anything any other agent would not know that and then quite frankly every proprietary platform has that advantage of being able to have content that you know that is not publicly available so i think the key for me is we've got to basically we've got to basically use the use the kind of you know the the enablers now to be able to get that moving faster but And ultimately, you want these platforms to be built, not having to depend on lots of specialists. So I think right now I'd call it a necessary evil.

Okay. Maybe this is a good time to ask about, you know, the value capture of AI, you know, how you think about a platform like Pega being able to monetize the functionality, you know, gaining the value from AI versus maybe passing on the benefit to the customers, or I guess even, you know, other layers of the stack accruing that value, like the model layer. Just how do you kind of think about what that looks like?

I think that there are, you know, there's three layers of value, right, that I think will synchronize as we become more mature. And those three layers are, there's the layer to the actual AI model providers. What value are they getting? Like, how much do they monetize? There's the layer of the PEGA level, right, the application, the platform, and how much do we monetize? And then there's the client layer, like how much does the client actually monetize out of this? And I think in order for that, you know, in order for this to be a sustainable model, there's going to have to be a value that's attributed appropriately to each of those three levels. There are things that you could do, like at the AI model level, you're going to see model selection, harnessing and managing the tokens. You're going to see that become increased. We're doing that right now on behalf of our clients, but you're going to see that be table stakes, right? People are going to want to know what it costs to run a model. They're going to want to know with 100% certainty that you always pick the right model. And that's where it becomes challenging when there's so many different models. So you almost need an intermediary in there to be able to manage the models and how do you manage that at companies, at large banks. I know that Citi cares a lot about creating this AI gateway where you're basically you are managing all of that by actually sending all transactions through your own gateway. I think that will become increasingly popular. And then I think that, you know, there's going to be this, you know, value attributing between the client and the platform provider, someone like Pega and someone like the client where, you know, we're going to have to give, you know, for what may have been the same cost of ownership in the past, a lower cost of ownership. The biggest place that's going to show its face, though, is going to be the time to go live, the cost to manage the operating system, the cost to upgrade. We want to eliminate as much as we can those costs and make these systems be able to build in an AI-native way that are upgradable, sustainable, that you can take innovation and evolution and manage security, vulnerability, risks, threats that come out. We really want to try to get that into a continuous flow as opposed to something that has these episodic updates that take, you know, tens of millions of dollars to go through.

I mean, you've also focused on driving predictable costs. Like, that was a big point of focus, it feels like, at the investor day was, or in your conference, was just emphasizing making it predictable for the clients. Just how critical is that for the customers and the types of use cases that you're focusing on? I guess, what does that also mean for, you know, balancing, you know, frontier model usage versus, you know, the open weight models or the open source models? Just how do you kind of think about what that means as well?

Well, I think that, listen, it doesn't, we all expected that the cost of executing AI was going to come down precipitously. We all kind of thought, oh, well, it's going to be, you know, the efficiencies are going to play out. The reality is they haven't. Why haven't they? They have, if you look at static use, but what's happened is the models have become more powerful. You have availability and capacity challenges, a lot of investment. So what's happened is for the same use case that you may have done in OpenAI 1.5, and now you're in Sonnet 4.7, the cost is exponential in terms of the amount of actual token use. You've got input, output tokens, reasoning tokens, right? Unestimable. You can't actually tell what it's going to take. The problem is not, that is a natural evolution, something that's not as powerful to something that's more powerful. The problem is you don't want to use the same model for everything. So I think there is a way that you can still welcome and embrace the most powerful models for the most critical things that you need to do. But for many of the AI use cases, you don't need the most powerful model. If you're trying to do an automated call wrap-up in a customer, for a customer service representative based on listening to a call and looking at all the notes and the transactions and put a two-paragraph summary together, you don't need the most powerful frontier model to do that, right? But if you're trying to measure the arc of a missile that actually gets launched in a combat sequence to be able to hit cert, okay, you're probably going to want the most performant. So that's one piece. The other piece is availability. If you look at AWS Bedrock, for example, they've got a number of models in there that they're constantly shifting usage depending on availability, performance, accessibility. They're actually, so that is the world we're moving into, which is right model, right time, supply demand, pressuring the whole system to get efficient on that. And I think that that was not like, I don't think any of us would have realized that like, we're conditioned that when, you know, Microsoft comes out with the 2006 version, there is no place for Windows 97, right? Like that's like, but in this world, there is a place, right? Because there's a different cost frontier for those different use cases. I think that's like a real aha moment for us. And we then use that behind the scenes to help manage token costs on Pegaside. And also, I think clients are recognizing that a lot, which is why each large client is picking their own models and negotiating their own deals and trying to manage the token usage to be the most efficient. Okay. Sense. Any questions in the audience?

Okay. Maybe I do want to ask maybe what this means for, I guess, the cost structure moving forward. If you're absorbing the cost here, how much of it is you absorbing it, taking a bit of a margin hit versus being able to route efficiently to the right model for the right use case?

So we give predictability to our clients. Predictability for AI does not mean free AI. It means predictable AI. So we actually tell you for a transaction, you will have a fixed cost on what that transaction And it's our job, Pega, to manage the token usage on the back end. And so what we're doing is we're taking away that ownership or that risk. We know best. There's nobody that knows best on which model to use at each step in the workflow at runtime than Pega does. So we actually understand, like, what are you doing? Why would you use which model? How do we want to manage routing to the right accessibility? So we take that on. Is there risk that what we charge as an uplift for a transaction might not be enough to cover Yes, that is a risk. Do we feel confident that that will not be a risk that represents itself 90% or even 5% of the time? Yeah, we feel confident there. But I just don't think that's something that a client is able to do because they would have to do that application by application. And so we feel like we're better suited to do that. So that's the way we're managing the cost of the client, which is fixing it and managing the cost to us by thinking about model choice, model selection, model availability, you know, kind of like, and also there's things you can do with the model. Like for example, imagine if ChatGPT on your phone, when you asked it a question and it gave you an answer and you asked the question again, I don't know if any of you realize this, it goes and re-reasons again to give you that answer. What if it just knew that you asked that question an hour before and repeated the question that it gave you? Imagine, like, those are simple things that you can actually do in the build of an application.

In the last minute here, maybe we could talk about free cash flow, everyone's favorite topic. You know, I think you've talked about maybe for the year coming in a little bit below the original 575 guide. You know, maybe what is that kind of, or what's the line of thinking around it now? Or, like, what's the right ballpark? And then, you know, as you think about the median term outlook you gave, the $700 million plus in a few years, you know, just what are kind of the puts and takes and the ability to kind of reach that number given the bit of a, you know, I guess, downtick here?

So, yeah. So free cash flow is critical for all businesses, certainly for ours, because it gives us all kinds of flexibility, buying back shares, like it just, you know, the capital allocation of value is huge. So very important to us. So two parts to that. One is, are ACV being lower? obviously we're not going to have a lot of time within the year to make up on any actual impact that has to free cash flow so there's some part of the free cash flow that's just you know given that there's only so much you could do within a year we are very confident that we will address any shortfalls we have there from a free cash flow standpoint so that 2027 and 2028 are on track to what we said. But so I would view 26 as kind of a year that we will make some adjustments to correct any shortfall that we have. But there's one thing that is very unique to 26 that I've not talked about a lot, and I think it's worth mentioning here, which is we settled our shareholder suit. We settled our derivative suit. We had these lawsuits that came out of out of the Appian verdict. And we are down really to the Appian retrial that happens in the first quarter of 2027. So our legal costs for this year are unusually high because it's just a lot of stuff coming. And a lot of that just relates to settlements or conclusions to that. So Steve, I did not add that back to our free cash flow when we guided. So we'll be very clear as we show how much of any cash flow deviation is related to that, because that's not really structural in the business. That's very much episodic. So I think there's, you know, we have to just, you know, our cash flow that we modeled, for example, was I think $30 million of legal costs for the whole year. And we might be $70 to $90 million for the year. So that's like, you know, that's not a small number that manages it. And that will not repeat, obviously, when we go into 2027 and 28 so that's those are the two components that we'll bridge back okay perfect uh i think we're at time so we'll leave it there but kenwin thank you so much for uh for joining us today so thank you thanks everybody