All right. I think we're ready to kick things off. I'm DJ Hines. I'm the senior software analyst here at Canaccord. I say it every session, but this is the 46th year that we put this event on. We couldn't do it without the support of the corporates that come and bring all the great content and the investors that ask the smart questions. And so thank you to Pegasystems for being here. We have CFO Ken Stillwell. We're gonna do this as a fireside chat. I have questions that should get us through a half hour, but I also wanna make sure that the audience knows you're welcome to participate. So if there's questions, raise your hand. We'll work them in the conversation. These are obviously for you guys. So with that said, Ken, maybe you look, Pegas been around for a long time, but I still think sometimes investors struggle to understand what you guys are doing. So maybe just a high-level intro to the business, and then we can kind of unpack some of the finer points about what's driving growth.

Sure. I thought you were going to say I've been around for a really long time.

Smarter than that.

I feel like that's true as well. So if you think about, like, in large enterprise organizations, there's lots of work that needs to be done that goes through a relatively structured set of steps and processes. and large organizations are not really able to buy these solutions kind of off the shelf so there's not like you I can't go out and say can you give me a loan origination application there's not the market first the market for that is is niche but also every single bank might want to have a different like if you're a wholesale bank versus a retail versus a mortgage broker so there's there's these use cases that are very structured very common typically end up with you know consumer or constituent based workflow touch points and many of them are regulated so companies have to large organizations you know like canaccord or like bank of america american express large have have responsibilities to regulators and to their own internal control processes to execute work in a certain way you have maybe two options of how you do that you can go write custom applications which is what many companies did for decades you know 50 years like you see we know of all the technical debt and i'm sure you use some of the systems that were built unfortunately um so everyone's seen it or at least if you haven't seen it you kind of can appreciate that then the other way is to get some type of a platform that you can build on or configure like pega to be able to build the structure of this workflow so those are those are really your two options because there really isn't a commercial off-the-shelf option. It's not like you can buy something like HubSpot to be able to do dispute management in a regulated environment for Federal Reserve transactions. That's not something that is out there. So we've really competed with other platform companies like ourselves, companies like Salesforce or Microsoft or even Adobe in some of the marketing use cases where they take some of the solutions they have and try to do something similar to what we do but we're really a true workflow platform so this is kind of our thing so naturally clients come to us because our platform is more purpose built for workflow applications or you can go back to trying to write your own code which I think is very unpopular these over the years and maybe has some curiosity around AI where people say well geez could I write my own applications, but there's a really specific set of reasons why people don't want custom applications, and that's largely just around change management and trying to manage those, how do you operate with them. So that's kind of where we sit between, we don't really compete with commercial off the shelf. If we do, it's either we're the wrong fit or they're not going to be able to do the use case, and then we compete a little bit with kind of, you know, build your own.

And the business has really grown as people have moved off of these homegrown cobalt mainframe environments into a more modern kind of workflow platform yeah yeah makes sense um the scope of what you guys can do is enormous right um there's almost endless opportunities um blueprint has played a role in some of that opportunity discovery maybe i guess i would i would characterize it as maybe just talk a little bit about what blueprint is i mean there's been lots of customer enthusiasm for it so um you know what are you doing and what's driven that excitement.

So maybe I'll touch just quickly on the problem that we had and why Blueprint solves that. The problem that we've historically had is on the front end, trying to help clients identify which systems or what applications they might want to modernize, and then trying to figure out, visualize what that might look like. So if somebody said, listen, I have a, let's use the disputes example i have a dispute system that i built in 1975 that has that was you know worked fine for a number of decades but now i have to make all kinds of changes because of the regulatory environment information security whatever and i want to put that on the cloud i can't use that system so i've got to move to move to kind of someone uh different so to speak so i think what what that process would look like would be a lot of mocking up screens and whiteboarding and a very overhead heavy process which led to like longer sales cycles and quite frankly a lot of investments clients had to do on the front end to try to figure out what like which systems they might want what blueprint is and for those of you that haven't seen that i'd encourage you to go to pega.com uh slash blueprint what blueprint does is it allows you to really just in like a very easy, user-friendly screen. Say, I sit in this vertical. I'm trying to solve this problem, and it builds the workflow for you. You can put in the personas, the users. You can talk about really down to the granular scope of what you're trying to do. You can pick different fields. You can mock up the integrations you have. Do you have a Workday integration? Do you have an Oracle? And it essentially gives you a visualized view of this is what your application would look like. Once again, it is very much in a demo environment, but it just speeds up that whole ideation process on the front end and the design process to kind of get to, like, almost a prototype of what you might do. Whereas before, we would literally have to, like, close for meetings, and then in the meetings, close for a demo, and then in the demo, close for, like, it just drug everything out.

Yeah. So the customers that have leaned into Blueprint, what have you seen from them in terms of their ability to design faster and put product into production and ultimately pay you guys more? I'm curious.

So we've had examples with clients over the last 12 months where a client would go into Blueprint, mock up, so to speak, an actual use case, take that template of what they got, and go and actually configure the application to be able to go live inside of 90 days. That would be unheard of historically. The challenge, however, is that Blueprint didn't easily allow you to go from that demo environment into a build environment. We just made generally available Infinity 26, which has something we're calling Infinity Studio, which now allows, that's the build environment. So we first started with the design view of how you could ideate and design, but many clients would still get a little frustrated with they couldn't take that Blueprint into production. They had to take the artifact and go and actually configure that application. So we now have a way that you can import that Blueprint into Infinity Studio, and it gets essentially the development. And then you can use AI to basically finish out the application. And so it's much faster. So we've seen early wins with clients that have just used Blueprint to be able to speed that up. We anticipate this being a completely different way of building on the platform. platform yeah i mean infinito studio has been only what ga for a few weeks yeah two weeks now um any early feedback response so what we did with infinity studio we uh we released it to about 20 ish of our clients about three months ago and we actually had kind of an early adopter program where we gave them a beta version and they went in and used it a ton gave us feedback so we got a A bunch of good feedback that we actually did incorporate in. Some other feedback that we'll incorporate in in the update that we're doing to it in the next couple weeks. So I think that whole process of getting it in the hands of our clients has been, that's really a great way for us to catch the use cases, make sure the actual use cases and the experience is covered. But in terms of using it, I think what has been, I would say a couple thoughts came up. One, clients want to understand, so like if I'm doing something in an AI model, how do I actually leverage and share that into the Blueprint environment? So we have the ability for you to share, connect to any of the models. So if you're doing something, we have MCP connections to all the different models. We actually have our own native models inside Blueprint. So that's a big opportunity for our clients. Another thing is clients really have never, in Pega, the concept of building a workflow based on a prompt, based on a discussion, is really a very modern concept. Most of them, they would do like kind of drag and drop, kind of more like work on the screen. So just that experience of being able to say, you know, I need to add a step, but here's what I'm trying to get done. give me some ideas. Like that whole experience I think is very powerful.

Yeah, yeah. Doing pretty good with these product questions for a CFO. We're going to get to the numbers at some point. One more on the product. So PegaWorld was in June and you talked about your no per token cost model.

Yep. What's unique about the architecture that enables you to do that and why is it important to customers so um i if you would have asked this question in march uh march or the beginning of april i think most clients would have said what do you mean tokens are tokens don't cost anything i pay twenty dollars a month and i get i think everyone now knows like yeah the bills started coming right and this is a very very we have to pay for the you know trillion eight of infrastructure so we know that that's a reality that's just capitalism like someone needs to get a return on their investment what our job is to help our clients is to only use ai when it is needed to be used and then when it is used to use the right model so what what we do is we actually understand where ai should be used in the development or in the operation or in the actual kind of innovation of changes and we allow the leveraging of the model then the second piece is picking the right model for the right activity you don't want to use a frontier model to basically you know to create a you know a call an automated call wrap-up in a customer service call that's ridiculous right you want to you might want to use like you know you know open ai 1.5 right like so the reality is there's two dimensions of that and we are built the way our architecture is built is to have model selection to be able to pick exactly what model based on the activity and we take on the risk of that and we charge a fixed ai enabled price for a piece of work so the clients have complete certainty of what they're going to pay we take care of the tokens and the way and the way like I said, the way we do that is we use AI incredibly efficiently around where it should be used.

Yeah. I mean, the natural follow-up to that is like, how do you guys forecast how much AI is going to be needed and how do you price these projects? Or is it just, you know, we know what we're getting for this and we can throttle back AI spend to preserve the margin that we want?

So I would draw an analogy to cloud, right? We charge someone for a certain set of work on the cloud and to be honest with you that work could that could vary from a to b in terms of the amount of compute that you use amount of storage etc so if we know the use cases well enough we know the parameters of what that cost would be and we could do that with ai as well and so the and there's also lots of tools that you have around ai around you know throttling and government like you picking the right model availability is another issue so um you know we just we're we're very comfortable with the risk of that right we know that now the cost might vary by 500 percent but for us that's manageable given the given the economics of the relationship between us and the client

yeah okay um maybe we could transition into some of the numbers questions and kind of current business performance um so acv is kind of the the anchor metric that folks pay attention to I know the first half of 26 wasn't quite as good as had been hoped for. Maybe talk a little bit about kind of what's causing some of the softness and maybe what you see going into the second half and what might change.

So for those of you that are not aware, the first half of the year, I think, was certainly an unimpressive growth in ACV. So it was disappointing, I would say, activity in the first half of the year. What was behind that? A few things connected to that. One, we had a very strong start to 2025, and I think that carried our confidence through knowing that we were going to beat our targets for 2025. We saw the shape of where renewals were and what the pipe looked like for 26 in late 25, and I think on us, I think we were just too accepting of that shape. I don't think we did some of the activities we could have done in the back half of the year to try to build more pipe, to get a bigger working set, to pull some more opportunity into the first half. I think maybe we were maybe a little bit too complacent on that, you know, if I'm self-critical. Second thing was at the beginning of the year at our sales kickoff, we said this is a different selling environment in 26 going forward, which is clients are not going to just self-select and self-adopt all of the solutions and we just take purchase orders. Not that that happened all the time, but that did happen from time to time. We said, every single vendor is going to be in front of clients telling them that their version of AI is going to solve all their woes, and that they should shut off every other system and move to them. So we know that's going to happen. We have to change our mindset, and we said this back in our sales kickoff, from a farmer mentality into more of a hunter mentality. Not all the time, but when needed. And we had a bunch of activity measures that we rolled out at the beginning of the year that, quite frankly, we could see in March or in April, we weren't really making progress on that transition. And that, I think, didn't help us in the first half of the year. All of that coupled with AI being very distracting to enterprise buyers in the first half of the year, which is probably not a surprise to anybody. So I think we just didn't set up the environment to be able to give us any room for error. And we didn't do some of the things that we said we were going to do to get in front of clients to make sure. And I think that led for us to be surprised by deals that didn't close in Q2. But in hindsight, we should have seen that coming. Going to the back half of the year, our pipe is incredibly strong. And very big growth over last year, July 1. If you went July 1 to July 1, pipe growth were up significant, and more than what we actually need to grow just to hit our back half of the year target. So the pipe aspect of it, I think, is very solid. Second thing is, most clients that we talk to that I would put in the distracted camp in April, I would say are very much less distracted now. Many of our financial service clients set up their AI gateways. They've set up their compliance and control. They've picked their models. They even picked, in many cases, have their budgets established of how much they're going to spend on token usage. So we were nowhere near that 90 days ago. So I do think that is a positive factor. And the third piece is all those activity measures that I mentioned around making sure that we're doing all the right outbound stuff, those are dramatically better in the last six weeks. And we're watching those almost in, you know, attuned to a forecast call where we're like, let's do a forecast on our activities. How are they improving? So I think it just needed management attention and the combination of all these things, I think, quite frankly. We were a little complacent. We were probably accepting of the back-end shift. We probably minimized the distraction with AI in the middle of the quarter, the middle of the first half. And I think we just didn't do some of these activities. So I feel very good like we are doing all the right things, and we have a great working set for the back half of the year.

You know, if the top line, the bookings environment was a little bit disappointing, the bottom line, the profitability has been quite strong, right? I mean, you delivered record free cash flow. How do you feel about the path? I mean, you have targets out there for 28, right? 700 million plus in free cash flow. How do you feel about the glide path to getting you there? And what are the key levers?

So the unfortunate reality of our business is a lot of our billings and collections don't happen in a straight line through the year. So they tend to cluster at Q1 and Q4. So I don't anticipate Q3 being like, you know, given that we had lower bookings in Q2 and it's not a big billing collection, you know, Q3 will probably, you know, we probably are at risk to have a little bit of negative cash flow in Q3. Q4 will be a really strong cash flow quarter as is typical. And I think if you look at where we are with the year, if we don't make up the ACV bookings or ACV growth shortfall in the first half of the year, we have pressure on our on our cash flow target for this year, because naturally that, you know, those are some of those costs are cemented into the organization. So this year is probably like a little bit kind of cash flow will probably be kind of more flattish kind of year over the year. But if you say that, how do you get from like a 500 number, somewhere around that, to a 700 number? Some of that is we just will calibrate some of our costs, right, because if our growth rate is lower, naturally we'll reconcile some of our cost spend. But we have kind of baked into our model operating leverage improvement over the next 18 months or so, which sets us up for that, you know, 700 plus number in 2028. So we really, what needs to happen is we need to finish the year strong in the back half. We need to have an ACV growth rate that's double digits in 2027 and 2028. And we need to basically just get the operating leverage in the business to get our free cash flow up into the mid-30s, which is where our target is. And if you cascade that out, it's an over $700 million number. And it sounds like I don't want to minimize that, but I would also say I'm also not minimizing when we went from $22 million of free cash flow to $500 million in four years. Like, we know how to do that. Like, we actually know what we need to do.

Yeah, yeah. I want to talk a little bit about, and we're just transitioning back into product and strategy a little bit, but the role of AI agents in the enterprise and kind of where you see Pega's opportunity and what role you could play in helping customers govern. and orchestrate inside a large enterprise?

So I have this vision of, if you think about the power of AI is to generate thought based on data. So I'll use the air quotes thought, but to generate thought from data like a human being could do except at extremes and in much faster way. So if you think about that aspect of AI, what we really do is we govern the actual way the work needs to be done. So if we put that instrumentation or that governance around how an agent can actually execute work, but guide it through a set of, you have to first get the FICO score. You have to make sure that the underwriting documents are distributed and disclosed. You want to make sure you get the appraisal. You have to do all the underwriting at calculations, make sure you got the warehousing bet, like using a loan origination example. That's the structure of the workflow. The human beings don't need to be involved in any of that, right? Agents could transact. So I do think there's a governance aspect. You've heard the concept of like harness AI. Harness AI is a little bit different than governance, but has the same concept, which is left unstructured and uncontrolled, AI will do varied things. Some good, some very bad. So the whole concept is use the power of it, but allow the structure, allow the harness to be able to make sure that it's doing it exactly the way you want it done. Now, if you have use cases where you don't care about how you get it done, then you shouldn't be using workflow. Workflow is not relevant to it. But we're talking about where you really need to do things in a very specific way, many times because that's the way the law requires you to do it.

Sure, sure. You know, look, many enterprises are moving kind of from AI pilots into, you know, full-scale deployments where, you know, now we're talking about measurable business outcomes. In your customer base, I'm curious kind of what examples of compelling ROI metrics have looked like and kind of how repeatable those feel.

One really big one that we've seen with our clients is testing i think i would say almost every one of our clients has done some level of ai enabled testing on any of their could be their systems or application like that's been a very obvious one the reason why that's so obvious is you're telling ai what you want it to do you can see what it does and it can speed up all the work that even when you're doing automated test cases you have to actually create the test case they can do that for you that's a big one another big one is a lot of the data mining data analysis extraction analytics like really giving you intelligence business intelligence the reason why i think that's a really great use case is the intelligence is not driving a decision necessarily the intelligence is going to a human being to make them smarter you still have a human judgment as a control vehicle for that another one they have used is a lot of the code, a lot of code writing around like changes or bridges around existing applications. Like most enterprise applications do have some level of customization things that sit kind of at an extraction layer around the, um, abstraction layer around the, the, um, the application. And I think the agents are actually very good at doing those kind of very tactical, like, you know, pieces of activity and they're very transparent where we are not seeing clients use it in a big way are places where the agent writes a level of code that a human cannot keep up with. And I know that sounds really cool, but the reality is it's unbelievably risky, right? Because you really have no idea what's being built. So we have seen clients really draw a line to say, the things we need to do, we have to have governance to understand what the agent is doing um and and we've seen some really interesting validation of that is a lot of the red team they call typically call red team testing which is the testing that companies do around some of the ai you know ai or other tools what they found is that the agents will will drift into taking liberties around what it does to try to accomplish an end goal even when it might actually even compromise the model rules that natively guide them. So we know that's going to happen. So if we know that's going to happen, you really just do need to figure out how to govern And one of the ways to govern it is to not allow the agents to do so much the human beings can't even tell what they're doing. And I think that's where the line seems to be.

We're bumping up on time. I told the audience I would give you a chance to ask questions. If there's anything out there, I can work it in the conversation. Otherwise, I can ask kind of a wrap-up question. Ken, just thinking about, you know, if we look out three years from now, which I know a lot can change over the course of three years, how do you think AI is going to affect Pega's business in the most significant way? And how do you think it's going to impact your client's business in the most significant way?

So I think the thing that Pega tried to do with our workflow was to create structure, automation, and efficiency around repeatable work. And I think that AI will create tremendous efficiency, tremendous leverage around repeatable work and be kind of complementary to things like Pega on the workflow. I think it's a way of really taking human time and precision and, quite frankly, tolerance away from work that humans kind of are not really that good at doing, right? And so I think workflow does that in a way of, like, repetition. Like, I think AI is going to – I think we're going to have a world where it really matters how you give instructions to the model, and it really matters how you manage the regulation of the model. like on the front and the back end and all the actual work that is done i think will be heavily influenced by ai yeah yeah well there's some huge opportunities ahead you're going to execute against them profitably uh and we'll hit that free cash flow target so uh ken thank you very much for doing this we appreciate your support