Moving past bolt-on AI to strategic reimagination - Interview with Don Schuerman of Pega

0:00So welcome to the next edition of the Punk CX podcast. With me today I have Don Schuerman, who is the CTO and Head of Marketing at PEGA. He's also an old friend of the podcast. We won't go into how many times he's been on the podcast, but he's going to get a jacket at some point, I think. We've got to find some arbitrary kind of like line in the sand. I'm excited for bright yellow jacket. Oh, there you go. It's laid down, laid down. And given that, it doesn't need much of an introduction, but I won't assume anything. And I want to just pass it over to you, Don, and say, hello, welcome to the podcast.

0:31Great to see you again. And is there anything you'd like to add? You want to say a few words about yourself? Yeah, sure. I, you know, like everybody now, I just do the same thing I did a year ago. I just stick the word agent into it, right? So I'm an agentic CTO. I head up agentic marketing for Pega. But in all seriousness, I continue to really be kind of lucky that I get to get at the intersection of technology and how our clients and enterprises are actually using it, which these days is a pretty exciting and fluid and rapidly changing space to be in. Yeah, absolutely. And I just saw a post that you put out recently about this kind of new superpower kind of model that you can't release into the world because it's just too superpower.

1:14So, you know, we can't really talk about that, can we? No, I feel like that's now the greatest marketing move ever from some of these AI companies. What I thought was just funny about it was, I think Anthropic actually very thoughtfully released this model and their project Glasswing to validate. And I think it's good that there's some responsibility in how these are being. But it was just OpenAI was so quickly behind with sort of like, oh, yeah, we also have a model that's too powerful for us to release. yeah it was it was it was very very quick jumping in there yeah it's like it's almost like playground stuff it's a bit like kind of like oh my dad's harder than your dad yeah exactly my model's also really far too dangerous for me to release but anyway that that's an aside now we are in april and we are quickly approaching pegaworld which is going to be held in vegas in from the june the 7th to the 9th getting excited I am getting excited.

2:14It's a little hard to get excited when earlier this week I had to drive to speak at a conference and I did so in a flurry of snow in Boston. But June is closer than we believe. So I'm starting to get very excited, especially I just we have a we have a really great spate of customers. we have our partners from AWS on stage our partners the CEO of Cognizant who is not only one of our great partners but also I think a really great thinker in terms of how AI is transforming businesses and so I think it's going to be a really interesting set of conversations I mean I'm well unfortunately I didn't make it last year but I was there the year before and thankfully I'm going to be able to make it this year but I wanted before we actually kind of go into what we what we're like to expect i wanted to just pick up on what you were talking about you talked about ai and agents and it all gets tacked onto everything right now and there's a lot of noise around all of this sort of stuff you know vendor headline promises and they're all sort of like shooting for some sort of nirvana of autonomy intelligence and transformation at scale i mean i want to understand what's actually going on in the real world and so in the middle of all the noise don can you tell me what actually is working in the real world with all this new technology because you know without doubt it offers incredible amounts of promise and potential but i want to understand how what you're seeing in terms of how it's being applied to digital transformation customer experience and driving those you know ever important kind of like outcomes like any big technology change right there is truth in the hype and then there is hype in the hype right and the models are getting phenomenally powerful we've been we've internally at pega you know have over the last year or so completely reset our engineering team and the way we do engineering now that we you know our engineers jobs is not to actually hand code much of our software is to to orchestrate and govern and manage the agents that are out there doing a lot of the coding so and i think that's going to continue to be true where i think where i think the challenge is and you know this is especially true in the cx side of the world is that real power runs up also against the real complexity and real legacy systems and real regulatory environments, which big enterprises need to engage with their customers.

4:50And, you know, I think the opportunity and the challenge is not the models themselves. You know, we're going to keep hearing they're going to we're going to keep having this sort of escalating model war as to who's got the most powerful models. Like we are barely scratching the surface of like what the models we have today could do inside an enterprise. Yeah. And, you know, what we're going to talk a lot about is how can you actually apply some of this power of AI first to reimagine how the business even runs, right? Because I don't think, I think we're still tending to think of customer experiences through the lens of a pre-AI world, right?

5:27So we need to almost rethink them. And then we have to figure out how to run them in a way with the right mix of AI and deterministic capabilities so that we're doing that in a way that is fast and predictable and scalable and doesn't cost us hundreds of millions of tokens across every single customer interaction, right? So I think getting that balance right, that re-imagining and that run is where we're really focused on when I'm talking to my clients. And I mean, because it feels like we're still in that kind of bolt-on era, where it's a case of like, I'm going to put a turbocharger on my car and the car's going to go faster and further and all that sort of stuff, where actually what you're kind of saying is we need to reimagine the car and a different way of traveling.

6:19Yeah, the metaphor of where I think we are is I kind of almost feel like, remember when websites first popped up, right? This is going to show my age here. But like when websites first popped up and enterprises began using websites as a way of engaging, they would have these customer service forms on the website. And behind the form, it would just email somebody. Yeah. Right. And I feel like we're almost kind of in that phase, but AI, right? We haven't, we haven't adapted, we haven't updated the process so that we can actually plug AI and use it and make it valuable and engage with these new ways that we're, we're still plugging the legacy way of working into a very modern and different experience.

7:04Yeah. And, and I think we need to, we need to be thinking about and doing the sort of hard work of alignment around what do we want those new experiences to be. yeah no i think that's kind of great it's a it's a it's a great point i mean i've seen kind of people you know there's increasingly a conversation that's starting to say well you mentioned alignment and actually achieving that alignment but then also kind of like pitching out okay what's the vision of how we want to you know to operate and kind of we reimagine kind of like how we kind of do business and how we engage and and things and and then and that gets kind of amped up i think when we start talking about agentic systems which is the other kind of kind of big hoopla sort of thing and and there's a lot of hype and promises are attached to that to those kind of words i mean i wanted just to get your take also on that sort of the whole agentic approach or pega's approach to agentic ai and and why it's different because i just want to just i know it's different and i know you've got a point of view on that and it's a very fixed kind of point of view.

8:09Well, not fixed point of view, but it's a very sturdy point of view. I don't think anybody can afford to have a fixed point of view these days, right? I think this is, we're in a world where we need to be adaptable. What I do think is this, I think the way in which we will actually drive a lot of agentic automation is on the back of still quite deterministic processes, right? If I'm a bank and I need to handle a complaint from a customer, or I need to manage a bad transaction or I need to handle a fraud investigation, the way in which I need to do that is pretty fixed. It's fixed by regulators.

8:42It's fixed by their best practices. And I can actually pretty well, you know, define it. So what I don't need is an agent rethinking it and re-reasoning it every time that introduced risk that introduces a whole bunch of cost. What I do want to be able to do is think about, OK, if my customers are increasingly coming in through agents, either by agent or Or maybe even my customer's own personal agent that is talking to my agent. Do I have the ability to discover and find exactly the right workflow or process that's going to support their needs and support their requests? So can I connect that to them instantly?

9:19And then as I go through that process and I go through that workflow, can I dispatch agents across it to do what historically might have been manual work, right? Right. So we're working with a couple of banks in this area of what we've called agentic complaints, where when they get a complaint in from a customer, they need to be able to, one, figure out what kind of complaint it is. Right. And how they handle a discrimination related complaint versus a fraud complaint versus just a bad service complaint are all actually different in terms of what they do and how they report it. So having an agent on the front end that can interact in a very conversational way with the customer, capture information.

10:02And then once it learns what the customer issue is, then it's got a library of workflows for EPEGA that all exposed sort of as a context file. Right. When and how each should be used. The agent can grab the right workflow. Oh, this is someone who's complaining because they didn't get along because they thought they were old. Like this is an age discrimination to plan. All right. How do we how do we now process that? Yeah. Right. But then across that complaint, you know, the bank might have an agent that they've already built that does fraud analysis. So we're going to take the claim information.

10:35We're going to throw that to that agent to do some fraud analysis. Maybe we've built an agent in Pega that knows how to do research of previous complaints and find out and say, boy, have we seen a pattern of this at this branch before? Is this something we need to actually investigate systematically? So we're going to dispatch an agent to do some of that research and come back with a summary. So before that claim even hits a human being, we've already analyzed it, researched it, offered a point of view. So the human is just there to govern. The human is just there to validate and say, yeah, that makes sense.

11:08Move on. Or, hey, I have a question about that. Send the agent back to go look at some additional information and see what it comes back with. And there's a massive opportunity there to both deliver a better experience to the customer, hey, I'm responding to your needs quickly. I'm doing it in a channel that's really conversational and empathetic. But at the same time, do it with much higher degrees of efficiency on the back end. So rewarding both for the customer and for the operational efficiency people at the bank. And is that what you mean when you talk about predictable AI built for the enterprise?

11:41Is that kind of essentially what you're talking about there? Yeah, I mean, that's the idea of we need these AI agents to be able to be dynamic and responsive and add massive amounts of automation, but they still need to operate predictably. Right. I still need to actually be able to prove that I did certain things. And much of the work that I want to do in the business, if I'm in a big enterprise, is still deterministic. deterministic. Now, there's work that's not, right? There's creative work. There's, you know, in fact, I think one of the reasons why agents are so powerful at coding is that work is often kind of very variable and it kind of weaves back in and in itself.

12:19And agents could be really good at doing that. But I still need a human being at the end to check code. I still need somebody to validate the work that pops out the other side. Are some of the things that those things that we talked about, are those going to be the things you're going to be drilling into as the big themes of Pegaworld. So I think some of the studies that I, the case studies that I talked about, you're going to hear at Pegaworld. You're obviously going to hear a lot about how to build these kind of predictable AI models. Okay. You're going to hear, you know, one of the things that we're going to talk about is this idea of the sculptor and the clockmaker, right?

12:54So the ability to have the creativity, the ideas that AI can generate, but then also be able to plug that into, there are certain times where I just need a tick, tick, tick, tick, tick, And I need a very finely precision engineered machine. And I want to be able to combine those capabilities together so that I get the AI creativity doing what it does best. And I get the sort of time, the clockmaker operating in areas where I need speed and precision and sort of rapid response. So so so we'll talk a little bit about that. And then the other big thing we're going to talk a lot about is this whole idea of reimagination.

13:30Right. And, you know, I think. There's a lot of excitement and interest around things like AI coding and we're using it. It's awesome. But those tools don't tell you what to code. Right, right. And they don't do the hard work of getting an IT person and three business stakeholders and two end users in a room and actually agreeing that this is what we want the experience to be. This is the outcome we're trying to drive. This is where we want to make improvements. This is the regulations we need to adhere to. That alignment work, AI can do a lot. That's what we've done with Blueprint. AI can do a lot to help you envision a different way of working.

14:15It can put that into visual forms that a business user as well as a developer can understand. So it's not just in code syntax. And it can help facilitate that alignment because I think that's the fundamentally really hard work that the business needs to do. It's not just turn on these AI models. It's how do we get a bunch of different stakeholders together to agree on what we want the future to look like? And how does that future deliver better outcomes for our customers, better outcomes for our business? Well, I guess, isn't it kind of funny in that human beings are, we like to think of them as being deterministic, but they're actually more probabilistic.

14:55Oh, yeah, they're far more probabilistic, Right. And the the the you know, the the act of alignment is is such a is such a tricky piece. You know, I like I think it's I think it's really interesting to hear these stories of people, you know, who like I took Claude and I went home over the weekend and I, you know, coded an ERP system. And yes, but how many other people did have did you have to check in on that ERP system with? How many people had to sign off on it? How many of it like like if you just isolate yourself in a room with some code, you can bang out a lot. You can do a lot. But in an enterprise, you actually need to do that across teams of people in different stakeholders in different groups and competing pressures and like and facilitating that conversation and actually getting to a transformative outcome.

15:42that's going to be I think what really where where the organizations that really actually get value from this in a massive way not just the yeah we're summarizing our emails but we're really transforming the experiences that we can give to our customers and we're setting a new bar and a new standard I think that's gonna that's that's really interesting I'm really looking forward to seeing how you kind of bring that to life because I think that's the the hard kind of work I mean dealing with all the the technology and the and the models and all the infrastructure and all these different sort of things it's like those are all the tools are there but actually they kind of it's the soft tissue stuff that takes a bit kind of like longer to um to to change and to learn well old dogs and learning new tricks and all that type of stuff and it's also getting it's also getting the architecture right you know i think the the other thing the other thing that i think is really important with with all of this is ai generates a lot of code but when you've got a lot of code working together.

16:41You know, there's just an article in the New York Times, but we're getting to the point that we are now, we already had more code than we could deal with. And now that problem is just compounding and compounding and compounding. And that's where the discipline of architecture comes in. Like, I actually need to think about how am I componentizing this? How am I organizing it? How am I building it for reuse? And so one of the reasons why we think using AI to generate first the model, the business process, the componentry of the business, what that does is that enforces an architecture right enforces a componentization in a way that is very visual and visible and everybody can see it and move it but you're getting implicit sustainability you're getting the you're getting the promise of hey not only can i deploy this now quickly right and i think the expectation is we're going to be able to deploy stuff pretty fast but then i need to be able to change it and evolve it and keep it moving it like it it doesn't stop A hundred percent.

17:37You know, it's like, cause there's no point in building a black box when they, when, and then something kind of changes for the new regulation or kind of new market dynamics. Or if somebody comes in and says, God forbid somebody comes in and says, what's the black box doing? Right? Like you, you, you, you need to, I want to be able to have the power of AI, but I don't want to create more black boxes. I actually want to create visible into what is happening behind the scenes. Absolutely. So any chance you could maybe give us a preview or an example of a brand that you think actually they're they're getting it right they're getting they're they're putting a lot of this into in into practice it just because a lot of the time this can feel like to some people listening can feel a little bit lost i guess and it can feel a little bit esoteric sort of thing until you actually kind of put it into a real life example say like this brand a was facing x challenge and they did y and then they achieved zed yeah i mean so so i think there's you know i think there's some of them, they're brands that you've heard from at at Pegaworld before.

18:37You know, I just heard a great story from I was down in Sydney for an event we did down there. And I heard again from National Australia Bank and how they're leveraging Pegas customer decision hub capability to really have the next best conversation with our client. And this is really about, you know, using I guess we could call it old fashioned AI, like machine learning statistical, you know, it was three years ago it was what we called AI. But you have machine learning models at massive scale and really massive speed. Like this is not, this is models that are retraining themselves almost instantly to get smarter and smarter about engaging their clients.

19:23Now, what's really interesting is, and you're going to hear about this at Pegaworld, we're starting to partner with them on how do I use agents to define and build and sustain those next best action strategies. Okay. Right. So like, you know, if you think about it, one of the hardest transitions when you go to a next best action where every conversation is one-to-one to the customer. Yeah. Right. You actually need a pretty sustainable library of offers and things you can say to the customer. Mm-hmm. Right. It's different that if you've broken your customers down into 10 segments and I need 10 offers, I need one for each segment.

20:04Well, if I get down to a segment of one, I need thousands of them to be able to have something that's relevant to everybody. Well, it turns out that, you know, the agentic stuff is actually really good at looking at the data that we have in the customer decision hub about what offers are working, what populations aren't being served, what exists already. and suggesting and saying, hey, by the way, this whole group of your audience doesn't have enough good offers for them. But based on your other offer catalog and what we know about them, if you created these 50 offers, you would increase your relevancy by 20%.

20:44Right, okay. Right? Do you want me to create those for you? And then it goes and it builds them into the system. A human validates them. We can even suggest treatments and copy and test some things. But now I've got a sort of, you know, what I would call almost clockmaker decision engine that's running predictably fast, repeatedly with a whole bunch of audit and transparency in how it runs. And this sort of sculptor agentic AI that's reforming and taking the information coming out of that, forming new offers, coming up with new ideas and feeding that right back in to my clock engine so that it keeps running better and better and better.

21:24So we're going to be talking and showcasing some of what that looks like at Pegworld, which I think is pretty cool. That is very cool. It's almost like continuous optimization on a massive scale. Yeah, it's continuous optimization with agents driving much of that and then huge amounts of flexibility to have less or more sort of human in the loop in that circle as you grow it. Excellent. You know, another another client that I think has just done some great stuff. We featured a video recently from a Belgian telecommunications company called Proximus. And, you know, you talk about the sort of reimagination story.

22:01Well, they took Blueprint. Took an old sort of really kind of rogue IT service system that they had built. OK, was no longer meeting the needs of the business. Yes. The IT team didn't want anything to do with it because they hadn't architected it. Right. But they took information from that system, plugged it into Blueprint, reimagined a lot of those customer service experiences. And in a couple of weeks were able to go live with a new cloud architected, stable environment that their users like so much. They give it a round of applause when they when they went live on the first day. Right. So like like that, that ability of of leaning in hard to reimagine a way of working, of bringing business and I.T.

22:52together, you know, in a really collaborative setting so that by the time you actually get something live, not only do you get it live faster, because frankly, I think speed is going to become table stakes. But not only do you get it live faster, the thing you bring live is something that all of your stakeholders are already bought into. Right. Yeah. Which means it's better. It is a better experience for the customers, for the employees, for the business. I mean, I love this. I've always liked the example of kind of Blueprint and how you can ingest all sorts of data. I loved the demo from last year.

23:27I think it was where there was a video of a credit management system that was run on COBOL or something. I ingested it and then reimagined across a modern kind of platform. and I just thought there's so much potential in being able to take these things that in many cases many people don't know how they work they just work or if they do break then it's hard to get the people with the skills to actually to fix them again again because they're so old but actually being able to take it so easily and modernize it onto and migrate it onto a um onto a new platform i just thought it was it opened up a realm of possibilities and so i thought that some of those sort of things that and it was great and so the idea that you could extend that and then and it it helps to your point reimagine and well understand but also reimagine some of this technology that many of these organizations kind of like kind of work on so that's kind of great really looking forward to that but here's a key question i've got to ask and that is do people have to be there to get involved at pegaworld 26 or is it i can are you i can they get involved with watch parties or can they live stream or can they do all that there's going to be there's going to be watch parties they're going to be live streams okay there's going to be you know we'll probably we'll probably also have some you know post pegaworld best of hits that you can see and you know find but like the the the thing though that But I and you probably actually experienced this last year.

24:59Right. You know, you can get on the live stream. You can watch stuff. It's not the same as sitting down and actually talking to people. Yeah. And, you know, I think this is an interesting moment in technology because I don't think anybody's got the answers yet. I don't think anybody's like, you know, everybody's, you know, publishing playbooks and design. But nobody's nobody's actually written the definitive playbook yet. Nobody actually figured out how to solve this problem. So I and I actually think the way these problems get solved and the way the opportunities to apply this technology really gets worked out is when you put a whole bunch of smart people.

25:42in a place together where they get some big ideas that inspire them. And then they have the ability to go grab a beer, grab a coffee, and sit next to somebody and poke them. What are you doing? And what are you doing? And why did you do that? And did this work? And we tried that and that didn't work. And that to me is the real magic of an event like Pega World. So while certainly we'll have great ways on social and live streams, et cetera, to watch the event, I would also really encourage people, if you're interested, not just in what Pega is doing, but in how some of the biggest brands of the world, people like Wells Fargo and Unum and Vodafone and Verizon and different government agencies are actually really leveraging this stuff.

26:29That's the experience to me that is so valuable. 100%. I mean, I think having watched it from afar and then having been there in person a few times, yeah you're absolutely right there's nothing like being there in you know in person and it because you know you as you say and i get a chance i'm going to go talk to lots of different sort of people you know other you know you know kind of brands kind of the executives like you're in like you like yourselves kind of like other analysts and stuff and you just get it's a it's a really it's a big old mind kind of like fest which then yeah leaves me needing to sit down for a couple of days afterwards because just a little bit all kind of like to distill but but if you can't make it if people can't make it then definitely check out the live stream and stuff but do try and if you if you're interested in i would say do definitely try and kind of like get yourself along yeah pegworld.com i think we're i think we're having actually a flash sale right now for tickets so uh awesome good time to come in the in the highlights we'll make sure we can like put links and all that sort of in the in the in the show notes and just make sure and encourage people to go and check it all out and sign up so that's it for my main questions all righty unless there's anything that i missed out that you want to highlight no i think you know i i think we've got we've got in addition to some amazing client stories we've got some really exciting product capabilities you know we're we found the power of blueprint to just be amazing right and we continue to just see massive adoption and what you're going to see at pegaworld is us taking that engine that can use AI and agents and best practices to reinvent and redesign and then actually build your business and just embedding it almost everywhere in the product.

28:12You know, not just in Pega Blueprint, which you see online right now, Pega.com slash blueprint, but actually like in the studio experience for how you build and evolve an app in Pega in this agentic marketing capability of how you like, you're going to see this expand across. And I think that, you know, again, when I look at where the next step of this transformation is, it's can organizations reimagine and redefine how their business runs to take advantage of all of this AI? And, you know, I'm excited if we can play a little role in helping them do that, because I think that's the unlock. Absolutely.

28:48Brilliant. So some quick fire questions to finish up, Don. Yeah. Now, I've been using this format for a while, but the first one is all about your best advice. And I want you to provide that by completing this sentence for me. And the sentence is this. If you want to improve your customer or your employee experience, Don says, do this. Complete that sentence. Pick one process or better yet, pick one outcome. Because processes in our business only exist because there's an outcome at the other end. Right. If you have a process that doesn't have an outcome, you probably shouldn't have the process.

29:22But like pick an outcome, like fix my complaint or open my new account or get me a home loan. and rethink it. Get a group of people together, get some IT people together, get some people who really understand the business together, put yourselves in a room, grab a tool like Blueprint, pega.com slash Blueprint, right? Rethink it, redesign it and really poke hard on what do I need to do? Where do agents play in this? How will my customer want to come into this if they've got their own agents who are doing work for them? But this can feel like a very big world And I find if you like isolate it down to a single process, a single outcome, you then start, you can do something meaningful in a week or two.

30:10And you also build the learning so that you get better because you're not going to get it perfect the first time. That's okay. Like move to the next one, move to the next one. Perfect. Yeah. And it's all about breaking those bonds of inertia and trying things and generating momentum. It doesn't matter how much momentum you generate. If you're moving at one mile an hour, it's kind of faster than zero miles now. That's it. That's right. Right. Brilliant. So second question, punk related one, obviously. What company or brand do you think takes a punk approach to the customer experience and why? Oh, and by the way, can you remember the brand that you picked last year?

30:44So I think the one that I picked last year was Nudie Jeans, my favorite Australian jeans company. No, was that two years ago? I might have picked them once. I can't remember, but the last year was Pick Up Music. Oh, Pick Up Music. That's right. Pick Up Music, my guitar stuff. All right. That's good. So so this year's my I'm going to be less obscure. I actually think Anthropic is doing a pretty impressive job in terms of building kind of a punk brand. Right. You know, because for an organization that in many ways is almost taking over everything in terms of, you know, how their eyes used in terms of they were really surgical and targeting the enterprise market.

31:27Mm hmm. They've also, you know, built a reputation as the brand who's actually thinking about the larger societal implications of this. Mm hmm. Their recent decision to kind of hold their latest model. Right. Like the the some of the writing that Dario has put out about about like both the positive but also negative potential end states that I can can lead to. So I think that they've done a really nice job and I really love their sort of positioning of themselves as the tool for problem solvers. Who doesn't want to be a problem solver, right? And then making it so that you can go solve problems.