PegaWorld 2025 Don Schuerman keynote: Beyond Resilience: Architecting Tomorrow's Enterprise

Always fun to get to follow Rob because people are instantly just sort of in creative and statistical mode, all going at once. You know, it's a reminder that I have an amazing job. I get to work with incredible tech. The stuff that Kerim showed you yesterday, the stuff that Rob and his team is working on. I get to work with amazing clients: Unilever, Vodafone, Rabobank, Verizon, the 350 orgs in this room, everybody watching online. I get to work with an incredible team at Pega, the product teams that build our products, the delivery teams that make you successful. Our client engagement and Go-to-Market teams. The incredible team that puts on this amazing event. And can I have a round of applause for everybody, the product teams, the event teams, the creative teams, the caterers, and of course, our incredible partners and sponsors who make this event possible and make you so successful. You know, there has been a lot of innovation, I think a lot of provocative topics discussed at this conference so far, but I think there's a question that we're avoiding. It's in the back of my mind. It's likely in the back of yours. And if we can create a little bit of a safe space, I would like to address the elephant in the room. All right? So here is the wildly important question that I have been asking myself.

Where the heck are the flying cars? Right? I mean, seriously. Now this is not The Jetsons, for copyright reasons. But for those of you younger than me, The Jetsons was a cartoon. It actually started airing in 1962, and in the cartoon, it took place in the future. They had flying cars. They flew around. It actually took place 100 years later, in 2062. So by my math, we are two thirds of the way to The Jetsons. We have technology the Jetsons could have never imagined, right? So why don't we have flying cars? Well, maybe it's cost. And I don't Google anything anymore. I ChatGPT things, I perplexity things. We need a good verb. We need a new verb. So I perplexitied what's available in the flying car market. And you can get a prototype flying car for $1.3 million, which is about 30 times what it would cost you to get a brand new state of the art EV. But if there was value here, the market generally finds a way to bring price down commensurate with value. So maybe there's other things going on. Maybe it's making it regulatory. We currently deal with 1.5 billion cars in the world. We also have a million bicycles, which makes me a little happier. But it's complicated to keep all those things kind of flowing. And that's dealing with like just roads where the paths are set. Now imagine those are up in the air and go left, right, sideways. There's no roads controlling them. It would be a regulatory mess. There would also be a lot of risk, right? It takes thousands of hours of training and experience to become a pilot. My daughter recently turned 16. She is slowly working on getting her driver's license, and my wife and I both have a preexisting anxiety about her getting in a car and driving to dance class, which is about a mile and a half from our house.

And I think the thought of her walking out the door of the house, getting into her flying car and going 1000 feet up in the air to head to dance class, would just cripple us with anxiety. And by the way, it wouldn't be a very practical way for her to get a mile and a half to dance class. Which may be the answer to the question for a lot of the ways in which we need to get around, flying cars might just not be that practical. They add risk, they add cost. So maybe the answer is we just don't need them. And I think sometimes when we are confronted by massive change and stare at the future of tech, we guess a little wrong about what it's going to look like. And we forget that the future of tech is things that solve real problems integrated into the real ways of working that we have today. And I feel like sometimes some of the visions that I hear in the market of AI agents and how we're going to use them in our business, feel a little bit like flying cars, not just because some marketers have created cartoons from them, but because they're not actually integrated into the way that we really are going to work. Now, I realize I brought up flying cars, but maybe it's the wrong question. Maybe the question when we are confronted with disruptive change, is to think about how that disruptive change integrates into the problems, the challenges, the opportunities and the way we work today. So what are the agents, the AI agents we can actually use today? Well, you've seen a couple of them. Blueprint. Both for workflow and for customer engagement. Show of hands. How many people have used Blueprint? Show of hands.

All right. How many people have used Blueprint in the last two weeks? Not bad, not bad. Call to action if you haven't. Get thee to a kiosk in the Innovation Hub. Open up your mobile phone. Open up your laptop. Go to pega.com/blueprint. Try it out. Because we are constantly adding new capabilities, we are putting AI agents like Rob described and like Kerim showed you behind the scenes to make Blueprint smarter and faster and better at helping you redesign and transform your businesses, your workflows, the way you engage your customers. It is the power of agentic and AI reasoning put to profound and pragmatic use. And the thing that I love about Blueprint, as someone who has been around the process world for a while, is it gets you thinking beyond the cowpath. It gets you thinking beyond what you do today. And I think that's really important because while workflows and enterprise apps aren't going away, they are about to change profoundly. And for one thing, they will not look like this. Now, just one second, because I'm going to get on my soapbox for a second here. Hold on. Thank you very much! Thank you!

This is how Rob Walker feels every day. Okay. Can we please stop modeling our end-to-end workflows in BPMN or Business Process Modeling Notation? It is a horrible, horrible way to model our enterprise workflows. It is. Thank you, it is. It is not intuitive. It is not something a business person can understand. I cannot look at this process and understand what its purpose is and what it means. Even the technical people who are trained specialists in BPMN get on LinkedIn and have debates about what the process flow actually means. And while I love standards, we're going to talk about standards that we use all the time. This standard was designed in 2014, and given the pace of technology change, we are not going to re-architect our business for the agentic future and the digital way in which we want to work. Using a standard that hasn't been updated in a decade. Off my soapbox. All right. Luckily, you can take this archaic legacy brittle BPMN, you can import it into Pega Blueprint and you can create something intuitive, kind of beautiful like this, that reflects your processes in a way that an IT person can understand. A business leader can understand. Even a customer could understand where they are in that process. These processes can be used across any channel, your digital channels, your own channels, other applications.

We'll talk about agentic front-ends for these processes. But I think even these are going to change because we'll always have a little bit of the human touch. I think the fact that we all come together in events like this means that there is still that need, that desire, that one for human touch. I hope there is. There will be more and more of our work that is automated. More and more of the steps in these processes are going to be done by agents. We're going to have document agents that we dispatch to read through our documents, pull out information, validate that they are what they say. We're going to have research agents that we ask to go out and either look at the internet, look at our own knowledge stores, look at specific knowledge, public knowledge directories so that they can come back and synthesize that information, either for other agents or for humans to use to drive their process. Some of those agents are going to live in Pega, in Infinity 25, we're going to have a new document agent capability that I think is going to change your minds about how you can do intelligent document processing directly inside of your Pega apps. But a lot of those agents are going to live outside of Pega, which is why Infinity 25 adds support for A2A, the Google agent to agent protocol, and MCP, the model context protocol that lets you set up all the different tools that an agent can use.

But those agents are going to be called inside of discrete workflow steps. And that's important because by calling them as a discrete step in the workflow, I can wrap the same governance around the agent that I already wrap around the humans who are doing many of these workflow steps. I can use skill based routing to make sure I'm getting it to the agent who has the right skills. If I need to put a quality check behind an individual step because the agent is new or the agent has been making mistakes, I can do automated quality checking if I need to escalate it because the agent is taking too long, or because human help is needed. I can do that. And if a human is involved, there will be agents sitting on the human shoulders helping them out. We call those coach agents, looking at the process data, looking at the broader context of what the person is trying to do and ensuring they have everything at their fingertips to get it done. And of course, the way we interact with these processes, with our workflows is going to change. Kerim showed you yesterday a Conversational Agent that used the language power of these large language models.

What we could technically call the semantic capability of the model to turn the workflow into something that we could interoperate with and that semantic power, let it both find the right workflow and then guide a user through it. What Kerim showed - I keep reflecting on it - was pretty amazing. If I had come up on this stage three years ago and told you that we were going to be able to take a video of a mainframe app, import it into a SaaS based tool available to anybody for free, that would then digest the details of that video, extract the workflows and the data models represented in that video, redesign the workflows in that video to take advantage of all the power of automation that we have today and ensure that they work across any channel or any conversation, lay out the data, models and interface points that that workflow needed to use, and then allow you to preview what that workflow would do in any application channel, both traditional UIs, traditional mobile, web, self-service, contact center, chat, and that you would be able to pick up your phone and call that workflow and talk to it in any language you want. And by the way, that whole thing would take about four minutes. I don't know, that to me is as cool if not cooler than a flying car. That is transformative technology.

And we were able to do that because we have the right architecture to make this happen. As we talk about this future with AI, I think it's tempting to think that maybe architecture doesn't matter anymore. That we're just going to throw large language models at the problem, and it's going to figure everything out. But I actually think as we move into a world where we are going to have interoperating agents and deterministic systems and humans all together, architecture matters more than ever before. And when we did the thinking around the Center-out architecture for Pega, we didn't know that agentic AI was coming. We had some ideas about AI, but we couldn't have foreseen, I don't know if anybody could have foreseen the wave of change and disruption that's hitting us, but we did know there were some principles that would help us get the architecture right.

If you could separate out your business logic, the decisions you want to make, the processes and workflows you want to run, the case structures that manage all that work. If you could separate that from where data was stored, you could insulate your users and your customers from the messy complexity of your back-end. You would ensure that that messy complexity didn't creep in to the efficient customer experiences that you were trying to deliver. So that virtual separation became really important. And now that virtual separation is allowing us to create the cloud native data stores that Alan talked about yesterday, that are going to allow you to move more and more of your system of record data up into the cloud, we hope at breathtaking speed. We also believe that you needed to separate that business logic from your front-end, so from your Pega front-end, from your mobile, so that you could use the same decision, the same workflow, regardless of channel. So you had consistent experiences and you weren't duplicating a bunch of coding. And by doing that, we've now allowed it for agents to plug in really, really seamlessly to invoke those workflows, to be guided by them, to participate with them.

And of course, as we've been thinking about architecture, we know that this is never going to happen across just one app. Your enterprises have hundreds of apps. You have workflows and dozens, if not hundreds of places, some of them in Pega, some of them, unfortunately, probably not in Pega. That's okay. Which is why we had designed the Process Fabric to start connecting all of those workflows across different systems, no matter where they live. And as agents come into play, and we need those workflows to orchestrate our agents to make sure that they're always following the best practices that we've set for our business. That Agentic Process Fabric that Alan talked about yesterday is going to ensure that this works no matter where those workflows live in the system.

Our friends at Gartner have been talking about this idea of business orchestration and automation technology, BOAT, and how what we had as BPM and RPA and all these different pieces is coming together to support a world where we are orchestrating decisions and integration to existing systems and humans and agents to get work done better and more effectively. And I think that's a great vision and a great way of looking at that, because I do think that orchestration and that power of having those workflows remains really, really important. So if I could I want to illustrate this point, and mainly just because it's freaking cool, I want to bring the drum bots back up here. So can I get the drum bots back up on stage? Drum bots, come on up. All right. They're not actually bots. They're just guys in suits with flashing lights. All right. Yeah, just letting you all know, just in case anybody was worried. But in this case, the drum bots for us are going to stand in for AI agents. And, you know, we talk a lot about sort of how we want to manage our agents and can we manage them with prompts.

So let's look at how that might happen. I'm going to give our drum bots a prompt, pretty simple basic prompt. I'm going to ask them to play some funky music. So let me ask this. That's the prompt I'm giving to our drum bot agents here. So let's see what happens. Drum bot number one. That was funky, right? All right. Drum bot number two. Wow. That was also funky. Very different from what drum bot number one played, though. Now let's look at drum bot number three. Wow. That was probably the funkiest of the bunch, but also still very different. Now, what happens if we ask all of these drum bots to try to play together? Ready? Drum bots! Ah! Stop, stop! That's enough, that's enough. They were funky on their own. But because they're all doing different stuff, it ends up creating a cacophony. Now, maybe we just need to do prompt engineering. We could refine our prompt a little bit. So I can make our prompt a little bit better. I can tell them to play some funky music in 4/4 time with a swing vibe and lots of sixteenth notes. So we've engineered our prompt.

We've made it more specific. Let's see what happens. Drum bot number one. Okay. Drum bot number two. Again, pretty funky, but they're still pretty different. Drum bot number three. All right. Now what happens when we put that all together? Drum bots go! All right, all right, all right. Thank you. That was better. That was better. But we're still not playing in unison. Now, we could continue to keep trying prompt engineering. We need to get some tools to help us do an automate prompt engineering. But here's the thing. We've already solved this problem. If you want to get a bunch of musicians to play the same thing together, you give them sheet music. You don't just give them free text prompts, you give them sheet music that lays out with room for interpretation what they want to play, but makes unambiguous the beats that everybody has to hit. And so, armed by sheet music, let's listen to our drum bots. And that is funky. Thank you, drum bots! If I could have drummers follow me around every day in my normal life, it would be about the coolest thing ever. Look, this musical notation, it's your workflow for rhythm. It ensures that everybody in your business is playing together, hitting the beats they need to hit in time with repeatability.

And Blueprint, as Alan talked about it yesterday, is the agentic composer. It helps you design the sheet music, the workflow that ensures your business works as one consistently to deliver on your customers. But I said that those workflows themselves are going to have agents embedded all throughout. So what does that look like? You may have noticed I came up on stage earlier and I was carrying a drum. And I actually have been experimenting with becoming a drummer myself. The drum bots are making that even more so. And show of hands. Is anybody here like a musician, either amateur or professional? Aspiring? All right. Anybody have amateur or professional aspiring musicians in your household? Okay. So I'm interested in learning drums. And I brought this up to my family, my wife, my kids, the niece who lives with us. And they said, that's great. You do you, we appreciate, you know, this is what happens when you get close to 50. But they also strongly suggested that I get a soundproof space to put those drums in. So I decided I needed to get a loan to do a little bit of home equity renovation, and I, like everybody at Pega, bank with this bank U+, which you may have noticed is our default demo bank here at Pega.

And so I opened up my U+ app and there was that conversational agent, connected to all of the U+ knowledge and workflows. And so I asked that agent: I would like to get a loan to redo, renovate my drum studio. And that agent went out through predictable AI to the Agentic Process Fabric. And it looked at all the apps that U+ has. It looked at the customer service app, it looked at the financial investigations app, it looked at the business lending app, it looked at the home lending app and consumer lending, and it found the right workflow for me. It semantically matched my request to the right workflow, which was this home equity workflow. And notice that we use the creativity of the semantic language to stick a little bit of brand in there. It gave me a little joke. It was a little bit humorous, but when it actually came to doing the work of my loan, we actually restricted the creativity by pointing it to the workflow that it's going to follow.

And I slowed this down to show you, but all of it happened in milliseconds. Now, armed with the right workflow, the conversational agent can go to work. So it can now start collecting the information that I need to get things done. It's going to start asking me for the key information like my name, my birth date, the last four digits of my Social Security number so it can confirm who it's talking to. It's going to ask me for some information about the loan I want to take. Like how much am I looking to borrow in order to do this? It's going to ask me to upload some documentation to help prove the collateral, like an appraisal document for my house. Now, like most of you I carry the appraisal document for my house around on my mobile phone at all times. So I uploaded it for the agent, and then the agent dispatched that to another agent to process it, to actually analyze the appraisal doc. Figure out one, was this a valid appraisal? Was it for the address that I'm trying to get an equity loan for? Does it sit in range of the value that I'm looking to ask. And it was able to do that. Got the appraisal document processed and kicked off. Now the rest of the workflow to go get me my loan.

But our conversational agent is also armed with knowledge. It's connected to Knowledge Buddy, a knowledge agent. You can check this out in the Innovation Hub. And that knowledge allowed it to remind me that I should look about permits. Probably good, since I'm about to disturb all my neighbors. So I asked it, what do I need to know about permits? And now the conversational agent went to the Knowledge Buddy our knowledge agent, and it got me an answer about permitting and noise laws in my area. It didn't just go make it up. It went to a collection of documents that U+ had provided and said, this is where you will answer the question. And when it answered the question, it actually came back not just with an answer, but a citation. It pointed me to the doc it got the information from so I could trust and validate that it was the right information. So in that conversation, I was working with a conversational agent to interact with, it went to the Agentic Process Fabric, it found the right workflow, it got guided through, it dispatched the document off to an automation agent to go process the document. It connected to a knowledge agent, Knowledge Buddy, to get the information that it wanted. So now let's switch hats.

Let's become the lone processor working in the operations side of of U+ Bank. So our lone processor is logged into her beautiful Constellation portal. And there has been some chatter in the U+ rumor mill that there are a lot of musicians asking for loans these days, and our loan processor wants to validate that, and there's no report. But rather than having to ask IT to create one, she just asks a free text question, what's the loan volume in the past 90 days to musicians? And we dispatch some AI to go write that report for her. So we literally just build a report on the fly based on her text and explained, yes, the rumors are right. There have been a lot of loans going to musicians. I get a lot of people were inspired by the drum bots, inspired by Sugar Ray and En Vogue and DMC, and they now want to form bands. They want to be pop-rock, so they're doing the renovations necessary. We could take this data and now use that as an input into some of the stuff Rob was talking about with Next Best Action and Customer Decision Hub to maybe start making some better offers and more tailored offers to these musicians, but for now we're just going to move on with our loan processor.

So she's going to go back to her home screen and a new request, new case for her to work is going to pop up. And it's my loan. Great. So she's going to open that up. And the first thing she's going to do is she's going to get a coach. This is an agent that's designed to guide and help humans so that coach goes and it analyzes the case. It analyzes the information about my loan. It also can automatically go do research. It can do Google searches or internet searches or look at sort of MLS data, which is data about homes that are being sold in the area. And the agent's noticed something that it wants to point out to the loan processor, which is, there's actually been a surge in value in homes that have home studios. So maybe that appraisal that Don was getting was wrong because if he's going to put a home studio in, the value of this house is going to skyrocket, everybody's going to want to be in the house that has a soundproof drum studio. Go figure. And it's recommending that we dispatch a new appraiser, but we find an appraiser that's a musician that really understands the coolness that Don is about to put into his house. Now, U+ does not have any existing workflows to dispatch drummer appraisers. Not a situation U+ had planned for. But that's okay, Because our loan processor has a new capability we put into Infinity 25, called an AI ad hoc workflow.

So what she can do is she can describe the work that she needs to get done into the system. She needs to dispatch an appraiser, preferably a drummer, to get that done. And then from that description, we're going to call out to those same design agents that power Pega Blueprint, but we're now going to do it at runtime. So we're going to literally design a new case type and the workflow that goes with it at runtime for our loan processor to run. But notice the agents aren't going to go off and just do this. They're waiting. They're waiting for the loan processor to validate. Yes, those are the right steps. We could even have this go to the loan processor's manager to validate that these are the right steps. She can actually edit, change it, add steps, remove steps. But once she agrees that that actually is the right new workflow, we've now created that workflow in real time using our design agents. She can dispatch it. And now Pega will kick it off.

Go dispatch the loan processor, the new appraiser. Go get all that handled so my loan gets taken care of. And the cool thing is, we can now take that workflow and if we want to establish it as a new canonical way that anytime we get a request for drummer appraisers, we'll just follow this process. So we use that whole design agent power, but we used it at runtime to build a new workflow just in time for the work that we needed. So what did we see? This is that kind of world of agents that Alan was talking about. We saw a design agent that both built this whole U+ home equity loan system. That was Blueprint. But also we saw one run at runtime to build me a new workflow to dispatch an appraiser. We had a conversational agent that I interacted with to get my request. We dispatched an automation agent to process my document. We interacted with a knowledge agent to tell me all about permits. We had a coach agent helping guide the loan processor to trigger the fact that maybe a new appraisal was necessary.

We used AI to write reports. This is a multi-agent world and armed by the Agentic Process Fabric and the core orchestration, the BOAT capabilities in Pega, we can orchestrate them so that they all follow the same rhythm. They all hit the same beats and I, the customer, am quickly and efficiently delivered the loan I need to build my cool new studio. And all of this stuff is real. You can see it in Blueprint. You can see it in the Innovation Hub. You'll see it in Infinity 25. And it takes the form of what I like to think of as integrated disruption, massive transformation. But that pragmatically works with the things we need to do today, the problems we need to solve today, the compliance and rules we still need to follow today because we might not have flying cars that have replaced all of our roads. But we did end up with drones. I can for 100 bucks on Amazon buy a drone that I can fly up in the air and take like studio quality crane shots in my own backyard. We can fly drones into caves to explore places on the planet we've never seen. We can accelerate search and rescue.

We have GPS in our cars and self-driving features that get us where we're going faster, easier, safer than ever before. We have a commercial aviation system enabled by these things like GPS and autopilot for planes that allowed all of us to come together from around the globe. And that allows all of us to move around the globe in ways that the Jetsons never thought possible in 1962. And my guess is, unless you came through Newark, you got here relatively conveniently. Transformation doesn't always look like the cartoon in our mind. It's integrated into existing tools and approaches, but it is often equally, if not more, profound. And that's what has allowed us to enter the autonomous enterprise. We now have agents that can design, automate, optimize all of our workflows and decisions that run our business and serve our customers. You've experienced this right here at PegaWorld. But we need this to go far beyond this room. So in the second half of the year, we're going to be taking this message to the whole world.

We're going to be at events, running events around the globe. We want you to bring yourself bring your colleagues so that we can tell this story so that, like Vodafone, we can all do "No sprint without a print", and we can deliver apps in 40 hours. We can all rethink all of that legacy debt that's slowing down innovation, because our mission as a company is to change the way the world builds software. And we want to do that for every enterprise that has the needs of more efficiency, removing their legacy debt, better serving and engaging their customers. We want to all work together to that vision, and I want you to do that. I want you to build lots of Blueprints over the next year, and then I want you to come back here and share what you achieved at PegaWorld 2026. So thank you very much! We will see you next year. Have a great wrap up of PegaWorld!