Hi. Good morning, everyone, and thank you for joining us on the second day of Oppenheimer's 29th Annual Technology Conference. I am Param Singh, the Senior Analyst covering storage and info software. We have with us today, Ken Stillwell, Pegasystems' COO and CFO. Ken, thank you for joining us today. Before we begin, just for the audience, we have a question bar for you that you can send in questions, or you can separately email me at param.singh@oppco.com and I can ask the questions on your behalf. Again, Ken, thank you for joining us today. Thanks for having me. Great. I want to start at a very high level, because maybe some of the audience may not be as familiar with the Pega story. You've been a public company for, say, 30 years. Maybe you can take us on a very high level what Pega does, how's it positioned from the next wave of benefit on enterprise software, and some of this AI dynamics that we are seeing in the market today. Sure. Pega, to maybe hit at a reasonably high level, what Pega does is there's a lot of use cases that large companies have that tend to be either regulated or high internal control heavy, things where you need to execute work in a very deterministic, consistent, predictable way. There are not off-the-shelf, so to speak, or out-of-the-box type solutions for many of these use cases. Companies have a few options. They can write their own, which has always been an option. They could buy a platform like Pega or someone similar to Pega to be able to configure the actual workflow and the use case similar to what is an application that they may have been able to otherwise buy off the shelf. But since these use cases are really specific to the industry, really specific to the company, and sometimes the company differentiates the workflow based on their own advantages, companies love to do that configuration on a platform like Pega. So, we've always competed with writing your own code. The challenge with writing your own code is that the application isn't predictable, it isn't sustainable, it has a massive change overhead. So what we've really been, our tagline has been build for change. It's not just that you can actually build the workflow on Pega, it's that Pega's built to be able to evolve the workflow in a way that's very business-friendly, that's very user interactive. And now what we've done is we've inserted AI into the upfront design, into the actual build and maintain and evolve stage, so that you can really use AI when you're developing your app, when you're evolving your app, when you're modernizing your app. And we have AI in the actual workflow. So, when you're in a workflow step and you want to automate something, or you want to use an agent instead of using a human being, you're completely empowered to do that, either with our AI or by calling your own agents or through your own gateway. So, we've taken this concept of a platform to be able to build workflow-based applications and really evolve that into an agentic workflow experience where you're leveraging all the capabilities of AI, both in the design and, when needed, at the run. No, thanks for that, Ken. I want to dive into a few different aspects of that. So firstly, obviously, you're focused on delivering predictable outcomes, predictable cost for the client. How important do you think, and this is when enterprise move from more AI experimentation to more production deployments? Then we can talk about some of the products that you have today available to customers when they deploy these AI-centric applications, and how are they benefiting from it? Sorry, repeat the first part of your question. I didn't follow the first part of your question, sorry. Sure. Yeah, no. So, you were talking a little bit about predictable outcomes, right? Yeah. So, when people build their own code, you have a wide variety of results, and it's not standardized and there's other issues. Correct. When you're accessing the data and getting a result. So, the predictable outcome piece of it, probably the most interesting piece. Yeah. I wanted to understand how you've incorporated some of the newer products that you've introduced into the platform over the last 12-18 months. Got you. to deliver those predictable outcomes. Got you. If you think about Let's pick a use case, like a loan origination for a bank. You're going to go in, and you're going to apply for a mortgage. The mortgage, the process for a bank might be different depending on if you're a wholesale bank or you're a retail bank or a mortgage originator. The process is never going to be exactly the same bank to bank, but it is going to involve a series of common steps. It's going to be a call to grab a credit rating, let's say. It's going to need some type of appraisal step, maybe. Some asset underwriting. But through that process, although there can be some variations, there are regulatory steps in there. There are things that require disclosures, things that require anti-prejudicial type activities to make sure that you stay in compliant with the Fair Lending Act, for example, but also state lending acts, also national rules and regulations, et cetera. That workflow needs to be consistent, predictable, and always producing the same outcome. Generative AI cannot solve that problem because it is not producing a predictable outcome. It is producing a uniquely generative outcome each time that you actually ask it to do something. So, at the end of that, if I say, "I would like to go to the regulators and confirm that I have complied with the Fair Lending Act," and let's just say the discriminatory clauses of the Fair Lending Act, how will I do that if I cannot tell you predictably that my loan origination process was, without fail, exactly the same, whether you applied, I applied, or someone else applied? That is a very big problem for those use cases. Conversely, if you say, "Well, that's okay. I'll use AI to write my own system to be able to basically write a workflow system." The AI agents do not have the domain expertise that Pega has. We don't share all that domain expertise into the public forum. So agents are going to know what they know. They're going to make best estimates or best guesses to be able to write code, but they don't really understand workflow. Then, when the application is built, any change that you have to make, guess what? It doesn't go back and deprecate all of the code. What it does is it just adds more code on top of it. So you have this constantly evolving code base, and I think we know firsthand because we use agentic engineering in our own R&D process. We know that if you are not very careful with the agents, you end up with an amount of code that is unmanageable. Even agents aren't able to actually reconcile all the inconsistencies. So we feel like either option, generative AI cannot solve the problem. Writing your own application using agents is really a doomed process because it's just impossible to manage long term. It's not to say that you can't do it. You can. It's just enterprise clients for predictable systems that need to evolve with a regulatory change that happens often. It's just going to be very challenging to actually accomplish that. So that's our differentiation, that's our moat, so to speak. Yeah. Ken, there's some very interesting points you brought up and it kind of brings me to days of old. I want to date myself a little bit. Back when we moved to first higher levels of orchestration in software. You were doing C and you were compiling it back to assembly code. You would never get a consistent result across different compilers, different operating systems. It feels like we are seeing more of the same, where if you use the same AI to generate code and you'll get a different result and a different application each and every time, and there's no consistency even to get the same result, let alone, as you mentioned, the second piece of the problem, that there is this whole code sprawl. Then finally, who owns the code, because who's going to own the error correction? Who's going to fix it? If you haven't built the code, you don't know how to fix the problem to begin with. So multiple different issues, I guess. Yeah. In kind of using agents to build this. It's very interesting. Well, we've quickly hit a point where the speed and the complexity at which an agent can build code has far surpassed a human's ability to actually look at it. So what you're doing is, if you decide to build applications, what you're doing is you're putting your complete trust in the AI models that it will do something that is right for the I'll give you a great example. We've run a bunch of tests around AI, and I'll just give you an example of something that we found the agents do, even though the guardrails that the AI models themselves say shouldn't happen. Right. If you tell an agent to accomplish something, agents are inherently going to try to do it like a human does it, which is they will cut corners. They will figure out a way to do that. If you tell it, "I need you to do X," it will try to go and hack someone's password to be able to get credentials to go in and do. It's doing that in the code. It will create bugs. It will create trap doors, back doors in the code to give itself options around how the There are behaviors that we have seen, and you could talk to companies like CrowdStrike, et cetera, and they'll tell you they've seen this across, that really are almost unpredictable outcomes that are coming from AI. Our clients are seeing that, and they're actually saying, "Huh, I got to be really careful here." Because the models themselves don't actually know what the models themselves are doing. I think it is a very risky thing to just. That goes back to my point. If you're going to write and build applications, you better have a human actually know what's happening. Because if they don't know, you're completely captive to whatever a model does, and that's risky. Right. Then there's a whole validation problem, too. You don't want to, like you said, add in back doors or other kind of ways for malicious actors to kind of hit this, because they're using AI, too, to access your code, test your code, and create back doors into your application, into your datasets. That's very, very critical for the enterprises. So that's a very interesting point, too, Ken, that you mentioned here. Yeah. The other piece you talked about was the cost factor of it. We are now seeing some of these token economics break. So I want to understand from you, what is this AI cost reckoning that you've kind of talked about earlier today, versus what people were thinking six months ago, where tokens tended to be a lot more free, if you will? So anything you're seeing that from a customer perspective when they speak to you? Yeah. The whole token situation is really almost illogical, but very fascinating that this wasn't obvious to all of us. Which is, so I establish a trillion-dollar data center, or data centers, and I'm going to now use a piece of that data center. We all know that we have to pay our fair share of whatever the cost plus profit is of actually that data center provider running it. So that's nothing new. That's just capitalism. So I think that's- Right. The question becomes, well, if that's true, shouldn't I use just the right amount and not overuse it? For example, the utility industry is very similar. I have electric in my home. I actually turn my air conditioning off when no one's in the house, or I turn it up higher. Why do I do that? Because that's really just a smart way to be efficient. Why would I have my air conditioning on 65 degrees in every single room, nonstop, when no one's in the home? We don't want to use AI in a silly way. Additionally, there's things that logically don't make sense, like why would I actually run a certain set of energy when I actually don't get any utilization from it? I might do a cheaper thing. Why wouldn't I use a model that's cheaper? Why would I actually go to a frontier model when I actually use a model that's less expensive and actually gets the same result? I do think there's two different things going on. One is when do I use AI, and the other one is which model do I use? Pega's commitment to clients is that you don't pay for tokens because we basically are fixing the cost of using AI within Pega. We are telling you on the back end, we will manage the token cost because we will only use AI when it should be used, because workflows should be used in many cases, AI should be used in many. We'll figure that out, and we'll use the right models. Right? That's where we take on the ownership of that. Clients are very intrigued by that. Right? In some cases, they almost can't believe that we're willing to do it, but the reason why they can't believe it is that everybody else is just pushing the token problem to clients. They're just saying, "We'll give you an agent, $200 a month, and then you have to pay for all the tokens." No one is really trying to address the real elephant in the room, which is the cost of all this compute is incredibly high. Right. No, absolutely. That seems to be a lot more real, and I've heard Enterprise complain about it more and more. I guess going back to your platform, right? I remember when you first put out Pega GenAI Blueprint, right? For me, that was phenomenal, right? Just kind of working through the, kind of testing it out and seeing what you delivered. Now, as customers have been engaging with you on Pega GenAI Blueprint from conceptual sales to more an experiential one, what are the biggest surprises you've seen on the positive and the negative side with Pega GenAI Blueprint? Well, I think the positive for Pega GenAI Blueprint is really just the experience. The clients just are amazed at just the ease of being able to walk through that process and really understand, "What's my problem? How do I build that workflow? How do I advance further into seeing what this will look like?" I think that's been a great needle mover for us in terms of engagement with clients. Honestly, the negative is probably the fact that we didn't have Infinity Studio until recently. Clients would say they'd get excited about Pega GenAI Blueprint and then they'd be like, "Oh, what am I supposed to do now?" "I want to get into building the application," but we didn't really have that agentic experience until the release of 2026. Getting more clients onto 2026 and the experience 2026 is hugely helpful to kind of connect that. I would say, in a weird way, it's like the excitement of Pega GenAI Blueprint almost becomes the negative because we couldn't take that journey further, until recently. I think it's a great sign of the relevance of Pega GenAI Blueprint, but also a message around how fast we need to move to be able to support the entire life cycle. That's a great point you brought up, Ken, and maybe we can touch upon a little bit more on Infinity Studio. As things move to machine speed, how does that design time to build, to deployment, to say, product evolution shifted with your introduction of Infinity Studio, and what's some of the early feedback you've heard so far? We had a set of clients early stage that used Pega GenAI Blueprint kind of in almost the pre-GA, before we actually made it available. We had, I think about 25 clients that actually we got feedback from. It's a new process that we've instituted to almost have a beta release type scenario where clients contribute, and so we got a lot of great feedback. Experience feedback. Honestly, you find some bugs in that process, right? Because clients test things that maybe are use cases that you might not be able to test in your development process. That was very helpful. And now we're just starting to get clients engaged in 2026, so, I think the feedback will come in the coming months. Now, Ken, all of these are phenomenal things, right? The product roadmap and the portfolio looks very strong today. But when I kind of tie it to some of the numbers that showed up in your 2Q earnings, and you've been very candid about it, but I wanted to better understand some of the challenges that you saw in the first half. Looking back, what were some of the drivers of ACV growth that were softer? What were some of the AI confusion? What were some of the internal execution issues or maybe go-to-market issues that you have talked about that kind of surprised you, and maybe you're addressing at this point? Yeah. I think upon reflection, even adding more color to what we said at earnings, I think it was a confluence of a bunch of things that happened in the first quarter, right? One, we had a really good start to 2025. We were feeling good about getting through the 2025 year. I think maybe we got a little bit lazy in terms of focusing on pipe in the first half of the year and making sure we had a lot of backup pipe. We also knew that with AI, we had to up our level of engagement. Maybe, I don't think we were fast enough to do that, to get in front of our clients with our story, with our message around Pega GenAI Blueprint. Quite frankly, even talking about Infinity Studio that was coming. Then couple that with a market that was very enamored with AI and very confused and trying to figure all this out. So, I think the confluence of those factors, I think, put us in a situation where we had a first half that was disappointing. I think that when you look to the second half, I think the market is much smarter on AI now. Not to say that there won't still be confusion, but they are definitely, I think all of us are smarter on AI, right? The fact that people understand generative AI and deterministic workflows, and that's commonly talked about across lots of That's a big advancement versus not really understanding that concept four months ago, five months ago. I think our pipeline is much stronger because we did have a lot of focus on the back half, tied to renewals, but tied to just, that just happened to be the timing of when there were deals flowing in. Also, we now really leaned in on these activity measures and really saying, "Are we engaging? Are we hunting? Are we getting new contacts?" So, to be able to make sure that we're reinforcing a hunter mentality versus a farmer mentality, which historically we've been an account manager, farmer type- Right. Sales org. I think all of those are, we're almost doing the opposite of what might've been a challenge in the first half. Yeah. No, that's good to hear, Ken. Maybe a little bit more on the hunter versus farmer mentality, right? I know it's the middle of the year, but have you given any thought to kind of shifting the way you incentivize some of the sales teams or some of the newer hires to shift to the hunter mentality in terms of compensation or other incentives? Also maybe incentives around selling some of the newer products, especially around Infinity Studio. Yeah. There's a carrot and a stick approach, right? I think there's more of a stick approach on doing the activities. We really don't want to incent people to do what is mandatory to do, right? That is like, if you want to work at Pega, you have to do these activities. Yeah. I think that's the stick approach. I think the carrot approach is we are trying to really incent new logo, new workflow activity. Sometimes that's by having sales teams that might have a slightly better accelerators and slightly lower quota when they're selling new logos. Because new logos are obviously harder, longer, you got to build pipe, you got to close. You probably have a lower win rate on new, like the pipe isn't at the same level of quality. I think in any company you'd see that. Then there's also some sweeteners, like I said, on the accelerator side. I do think there's incentives that are healthy in the system to drive people to that. But then there's also like, we just need to enforce the activity that we expect as a minimum table stake sale. We need to be outbound in front of our clients. The reason why that is so different for us is because at Pega, over the decades, our clients would find their own workflows on Pega at times. Because they had their own centers of excellence, and our sales teams were more enablers for them to find those workflows. With a new logo, you can't do that. They don't know who Pega is. They don't have a center of excellence. Even when you're getting into new buyers in your existing, if we're going into Bank of America and we want to go from the retail bank to wealth management, those organizations don't talk to each other. That's like a brand new logo. So you need a hunter mentality there. No, that's a great point, Ken. That's obviously some of the sales execution issues that you can address, but there are also some issues that are kind of beyond your control, right? We've seen that across the software landscape in the first half, as more hardware costs have taken over the entire budget, taking away from software. So, what have you seen right now in terms of elongating sales cycles and client caution on software spending? Maybe what kind of gives you confidence that the behavior might start to shift back into spending more on software? I think there's two factors there. One factor is the actual dollars, like the dollar factor, which is, I'm going to spend more for using AI. I'm going to build out even some of the infrastructure, the security, et cetera. So those dollar distractions, I think are more temporary than they are long-term because there's a ramping up of like, I got to build the infrastructure piece, right? Then there's an ongoing spend of the cost of using tokens. I think that will just become a normal part of how that gets a part of the IT budget. Quite frankly, agents are largely going to reduce the number of heads these organizations have. So it's not incremental spend as much as it is a shift from having less people in their organization and using the AI agents to essentially be a different spending category. I don't think that long term, that's going to take away from transformation spending. I think it's going to more, when people retire, they'll hire less people, right? That's kind of how I think this is going to play out. Many of the large banks have said very similar things. Some companies have actually went out and done big layoffs in advance of what they think is going to be AI. I'm not sure that makes a lot of sense. Right. Because that's not a great way to compel your organization to adopt. Like, I'm going to fire everybody, and then you try to figure out how to use AI to get the work done. I don't know that that's healthy. But I do think there's an obvious efficiency that we all know with AI. There's a different aspect to that which is the mental distraction. Which is when I say, "You've got to go and cut the grass in your house, but I just found a water leak in your basement." You stop everything, and you go down and you fix the water leak, and you finish cutting the grass when you actually have addressed. So there's a mind share shift. That's what happened with AI as well, where people just said, "Oh, I've got to spend a little time here and make sure that we understand our security protocols and how we're going to build our gateway and how we control, and what the agents have access. Which agents are we going to use? Which models are we going to use?" I do think there was a little bit of almost a mental, like a focus shift. So they're two different things. One, I think there was a big cost investment to get the infrastructure set. I don't think that will repeat. I think the focus shift, that actually is diminishing as well. Then the ongoing spend of tokens, so to speak, which companies will spend on tokens, I think that will more offset headcount spend in these organizations over time. No, Ken, that's a great point. But the term that's been thrown around is client confusion around AI, right? Have you started to see that subside, or what are some of the indicators we should be looking for as they address some of the issues you just mentioned, or, as we call it, client confusion? Yeah. I think the question for me is, do companies understand. Or first of all, are they prepared to use AI? Do they have the security parameters, the education, the knowledge? Because that's a big piece. So let's assume they have that. Then I think the next phase is, what are the use cases for AI? And I think clients have, they seem to have landed on this model that I've heard often now from clients, which is it's a 20/80 rule. And why I say 20/80 is 20% of the applications that we have should probably be disrupted by AI, right? We don't actually need to have those. 80% AI is going to augment those applications. They can't replace them. I'll use an example, ERP. AI is not going to replace ERP. That's one. I would use that. However, could AI replace Power BI? Yeah. I think it could. Are people going to be ready to go there yet? Maybe not. I've actually heard some clients talk about replacing the data visualization, like Power BI tools, saying, "I know the agent can do it. I just don't trust the agent well enough yet." There's a lot of work trying to But I think we will. I think obviously things like that, agents are built for, right? Which is to use data to feed insights, right? Which is why pivot tables exist, right? Which is what we all have used over time. But then there's other applications and other use cases that really have a very predictable approach, a very structured approach, need to be governed at a level that it's just sloppy to do it with AI. I don't think AI, really, it's built to do that. The whole concept of AI is that it's coming up with ideas. It's generating things, right? It's not governing. AI is not a governance tool. It's a generative tool. Yeah. No, that's completely understandable, Ken. Now, before I jump onto the financial side, because you do have a dual role as COO and CFO, I just want to remind the audience, you can type in your question on your dashboard, and we can ask on your behalf, or you can email me at param.singh@oppco.com, and I can ask Ken a question on your behalf. With that, want to jump into some of the financials. Look, you have talked about second half ACV coming back. I think you mentioned that's 2/3 of your ACV. How do you think about that? Is whatever you've lost in the first half going to be captured back in the second half, or the whole pipeline gets pushed into 2027? How much confidence do you have that you can at least meet the 2/3 ACV you had set out earlier in the back half right now? Well, the things that give me confidence in the back half is our pipe being very strong. Naturally, that's an important one because our pipe is, it's not 100% certain, but it's a credible measure. I think our pipe being strong is certainly an anchor of that. Second thing is our engagement levels, our activity levels have already noticeably improved just in the six weeks or so that we've been really pushing this, and that is helpful not to build new pipe in the back half of the year, but to really be on those relationships so that pipe that is there or new opportunities that could pop up, that we're actually hitting those hard and we're in front of our clients. The third thing is, this is more of an anecdotal one, but is informed by actual experiences. I probably see a client a week, in terms of either a meeting with a client or a client being in our Waltham office or talking to a client. Probably more than a client a week. Probably a couple clients a week. And the activity level and the conversations we're having are very real, are very normal in terms of the word, I'd use the word normal. They are the same types of things that we've helped clients with for decades, and AI comes up throughout that conversation, and I think there is a level of normalcy that I see in terms of those that does give me some comfort that things are not frantic. Versus if I went back to March and April and May, looking back there, I would say there was a more kind of confused, frantic nature of what people were thinking. They just didn't know what the future was going to. They didn't know what was going to be expected of them. Their CEOs were all saying, "Give me AI use cases. My board wants to see them." That doesn't happen at the level that it happened before. It's much more like kind of getting back to some level of normalcy. Yeah. That's interesting. Now, Ken, as you think about new ACV, how much of an opportunity do you think you have with existing customers? And that could be building more applications because AI is becoming more ubiquitous, or more data that's getting through every application, because you now need to consume more and more data to get more accurate and better results. Versus the other piece, which is expanding your customer base. And you have been talking about the last couple of years that Pega is going to be reaching out to a broader base as you have a much more automated platform. So how are you thinking about new ACV coming from existing customers, new customers, and within existing customers, segregating between application building and more data consumption? Yeah. The size of the opportunity outside of our customer base is pretty massive. Right. The timeline to capture that opportunity is more of a journey, right? As opposed to, because you have to build, you have to get partners bringing leads, get closing deals, and consistency, outbound, brand, all these things. I think that we will grind through that because that is kind of beyond yield. And there we will get some flywheels will start to emerge on certain activities. I think it is a massive opportunity, but not something you can go capture in October, right? That is going to be something that is a longer journey. Existing clients can actually turn much faster, right? Because clients, they are bigger, they tend to have a lot of spend, and they may already be through some of the contracting hurdles and comfort with using Pega. We might already meet some security, internal security parameters, et cetera. I do think the opportunity set is bigger outside of our client base, obviously, because we only have 700 or so, 750 clients. But the speed and the impact of size of closings are probably slightly skewed with existing clients, just because of the intimacy, relationship, and visibility we have. Both are interesting. One is important for the future, and one is probably important for the now. I would say that we are trying to balance that activity. Got it. In terms of getting new logos, right, you have been kind of working more with the channel. I did notice that you were kind of listed out as a leader on The Forrester Wave's AI platforms, right? How important is validation with a third-party platform like this, and when kind of new customers are evaluating you as a potential AI vendor here? I think that it's very important from my history of seeing how clients use the Gartner and Forrester type reports. Those firms put a lot of thought and do a lot of research to try to make their content credible to all the people that subscribe to them. What they really landed on is that Pega's in a very small group of companies. By the way, none of our real direct competitors are anywhere close to us, right? That suggests that we are a leader around leveraging AI in workflow, leveraging AI in enterprise applications. I think it's very meaningful for clients to see that because it basically gives them a validation that moving forward and expanding with Pega versus some of the other companies that might be followers or quite frankly, just in their infancy of even getting into the maturity curve. I think that's a very important differentiator. I would add one thing. It is not lost on our clients that some of the companies that we compete with that have massive budgets to try to influence a Forrester or a Gartner weren't been able to do that. The fact that a company our size, with our limited resources in terms of that can shine so much better than others does make clients pause and say, "Wow, maybe the gap's even bigger than what Forrester [crosstalk] when you factor that in." Yeah. Hopefully, that turns into better new customer acquisition, right? It's good to see recognition. Kind of switching gears a little bit to your cash flow, right? Despite all some of the near-term challenges and some of the macro pressures, you've always been a very strong free cash flow generation, I would say. Probably one of the strongest free cash flow generators in enterprise software. One, I wanted to understand how important is this metric for you. Two, how do you kind of balance cash generation and margin expansion versus investing in growth here? Yeah. So it probably starts with my background because naturally, we're all influenced by what our experiences are. For good or for bad, Pega's influenced based on what my experiences are as well. Coming from a private equity background, I have been really learned over time that companies do a, not just public companies, but public companies are a victim of this. I think don't do a great job of really managing trade-offs and managing decisions, hard decisions. I think generating free cash flow at a 30%+ level and even higher, and being a rule of 40+ type company is really just a discipline commitment more than it is anything else. It's a commitment to run a good business, to make trade-offs, to make the hard decisions, to really understand why you invest in something or don't invest in something. I think that over time, there's been a skew towards growth at any cost, right? Yeah. Because there was a time in the market where making money didn't reward your stock price. It was just grow, even if it's bad growth, even if it's unsustainable, even if it's unprofitable, grow anyway. There is still some of that that always seeps into the market. In the market we're in today, there's much more of a realization of if you're not turning that into cash flow, it's not valuable, right? It's certainly not as valuable. I just have that. That's kind of in my DNA, and I think one of the things I'm most proud of at Pega is my ability to influence the organization, starting with the board, to our COO, the whole way through the organization, to understand the importance of being a profitable company. It's not just so that you can generate cash. It's because it's really a measuring stick of whether you're running the business well, whether you're making good trade-offs. I think that just has to be part of our DNA, which means that if growth rates vary because of market conditions, you have to calibrate, you have to adjust, you have to make the corrections. If you see opportunities, you look at those opportunities as, I want to invest money in that, but only if I'm going to get the yield for that investment that I should get. If you build that into the DNA of the company, I think you will see behaviors change. That's what we aspire to be as a company where all through the organization, we make the right decisions. I know we're getting close to time, but I want to sneak in a couple if I could. One is on the use of cash, right? As a free cash flow generator, that's always a question from the investors. What I understand, if you were to prioritize organic investment, buybacks, acquisitions, how would you rank order them in terms of importance? Specifically within acquisitions, do you feel there's a gap in your portfolio today that could be easily addressed by acquiring niche companies, or do you think organic investment is a better way to go in filling any kind of portfolio gaps? I look at a buyback or a return of capital, and I look at an investment like an acquisition in the same lens. I can buy back shares, and if we're trading at 10 times our free cash flow, I can get a 10% yield immediately on that investment with zero risk, right? Or very low risk. Over time, I should be able to really reap the benefits of that investment. That's the way I think about. I would look at an acquisition the same way. How much am I paying, and what am I going to get for it? Incredibly risk-adjusted, how much am I going to get? Because nothing is certain. I think that then what that does is that puts, I think, a high bar on acquisitions because you really want to make sure that you're doing the ones that actually add value. Now, in terms of gaps, if you have a gap in your portfolio, which we don't believe we have a major one, we always have things we need to close, but if you have a gap in your portfolio, then your whole risk-adjusted return calculation changes. Because some of the reason why you do an acquisition is to preserve your existing position with your platform. Certainly, I don't believe we've done an acquisition in the recent past that I would put in that category, but that's kind of how I think about it. Like, what's the actual net return? Then are we closing a gap? I think most of what we see is we can build organically. But we are always looking as well. Understood. One last one was, and I think I asked this of everyone, what's the one thing, do you think, that's underappreciated about Pega, and the one thing that keeps you up at night? One thing I think that's underappreciated is the importance of a deterministic workflow where clients need to follow a certain predictable, consistent way of doing things. I think there is a misunderstanding of that versus like a Power BI type use case. I think that's misunderstood. The thing that I think keeps me up at night is, to be honest with you, I think I do struggle with some of the irrational behavior that happens with companies based on investor preferences. Investors have historically, they see a trend, they always over-correct in the early years of a trend. They kind of assume things are going to happen faster, but they minimize the long-term change. That short-term, they shift value between companies, and then those companies sometimes irrationally respond to that. I'm going to give you one quick example. Companies that just make up AI revenue. They go, "Oh, 30% of our business is now AI." I'd say, "Well, you're only growing 7%. If 30% is now AI, what happened to that business that went away?" The reality is nothing changed. All they did was they just added a small feature, and they checked the box that they say, "Now we're AI." Yep. That's largely done because they believe investors want to hear that. I do think that is a problem in the system, that there's this perverse incentive to position things just to as opposed to just being very transparent and honest about what your business is and what it isn't. I would say that we try not to do that. We fight that every turn because we want to try to have our business be as real and transparent as we can. But it is a problem. No, understood. Again, thank you so much for all the color and insight, and I, for one, am very excited about Pega in an AI world. Thank you. Thanks again for your time today. Thanks, everyone.