Scaling AI Responsibly with Pega: Driving Real Customer Results - Six Five On The Road

Step into the future of AI and customer experience, live from PegaWorld 2025!🐧

Host Keith Kirkpatrick, Research Director at The Futurum Group sits down with two brilliant minds from Pega: Tara DeZao, Sr. Director of Product Marketing, and Peter van der Putten, Director AI Lab and Lead Scientist on this episode of Six Five On The Road. Their conversation focuses on the future of AI, its impact on customer experience, and how brands can scale AI responsibly and intelligently to meet their business outcomes. #sponsored

Key takeaways from their discussion:

🔹AI's Next Five Years: Peter van der Putten outlines the significant key trends in AI that brands must monitor over the coming half-decade, painting a picture of the future autonomous enterprise.

🔹Driving Tangible Customer Experience Results with AI: Tara DeZao reveals how AI is currently impacting CX, distinguishing factors between brands achieving real results and those falling short, and the importance of starting with business outcomes.

🔹Navigating AI Implementation Pitfalls: Learn about common challenges in AI adoption, such as data quality and governance, and discover Pega's strategies to help brands effectively overcome these hurdles.

🔹Designing AI-Powered CX: Get an introduction to Pega Blueprint, exploring its unique capabilities for designing effective, AI-powered customer experiences and its role in continuous improvement and regulated AI deployment.

‍

Learn more at Pega and what’s next at PegaWorld 2026.

‍

Watch the full video at Six Five Media, and be sure to subscribe to our YouTube channel, so you never miss an episode.

‍

Or listen to the audio here:

‍

‍

Disclaimer: Six Five On The Road is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.

‍

Keith Kirkpatrick: Hi, welcome to Six Five On the Road. I'm Keith Kirkpatrick, Research Director with The Futurum Group and welcome to PegaWorld 2025. Today I'm joined by two of our guests.

‍

Tara DeZao: Hi, I'm Tara DeZao, senior director of one to one Customer Engagement for Product Marketing.

‍

Peter van der Putten: And I'm Peter van der Putten. I'm a director of Pegas AI Lab and Pega's lead scientist.

‍

Keith Kirkpatrick: Well, thanks both of you for joining me here today. So we were talking last year. So much has changed since we were here a year ago. Obviously we're talking a lot about AI agents. But before we even get into that and talking about all the great things that PEGA is doing, why don't we just start off by talking about sort of where we are going with AI? Peter, maybe you could just give a little color in terms of where we are and where we're going.

‍

Peter van der Putten:  Well, a big question, of course. You know, one of the things that I think we have as an overarching theme that should be a little bit longer lasting than just the latest AI hype is this idea of the autonomous enterprise, right? It says like, hey, if you can have autonomous cars where the driver's in charge, you give the car your goals, where you want to go, the car actually gets you there, gets you to work for you, right? Why can't we have autonomous enterprises that do the same, that apply AI and automation to self optimize towards certain business goals? And I think that's pretty much kind of an overarching vision that is still, I think, relevant, maybe even more so. You know, like if you then take in some of these latest developments like AI agents, it's very much geared towards how we can have an AI that plays a somewhat more active role, right? So rather than being some kind of passive AI that we need to write prompts to, it's an AI that gets to work for us and gets us to our destination, right?

‍

Keith Kirkpatrick: Well, one of the other things though that I think is particularly important now is we're hearing a lot about having more context to kind of drive these decisions that go on with AI. So it's not sort of like if you think back, you know, five or 10 years ago where you had, you would interact with the system and you would get a very static response. Today it seems like it is all about context and relevancy.

‍

Peter van der Putten: Yeah, yeah, absolutely. And then giving a bit more responsibility to the AI. I can give you some form of instruction. But you know, it could be, you know, you can count it on being vague, not like a very exact prompt, but just a high level goal that I want to achieve. And then you need to sort it out. So. And of course, agentic AI is capable to tap into like you give it various kind of knowledge sources or data sources or even actions it could take and it can figure out to kind of first collect more context to then figure out what it is that it needs to do.

‍

Keith Kirkpatrick: Right. Right. Now, Tara, you know, this all sounds great, but it isn't quite that simple when you actually deploy it within the enterprise. Right. I mean, there's other things that can, that organizations really need to consider. Correct?

‍

Tara DeZao: Well, you know, clients are at a different stage of maturity. Right. So we have some clients that are fully autonomous or on their way to being fully autonomous, and then we have some that are still stuck in traditional manual processes. So it really, you know, depends on the pace of transformation of the business. Right, okay.

‍

Keith Kirkpatrick: Right. So if you're a company that perhaps hasn't been sort of on the cutting edge, what's sort of the first thing that organizations need to consider? Is it their data, Is it their people, Is it processes or a combination thereof?

‍

Tara DeZao: It's a combination. I would say that what you want to consider first is what outcomes you're trying to drive to for the business and then work backwards. Data is foundational to AI. So, you know, that would be for me the thing you should focus on after, you know, the business outcome though.

‍

Peter van der Putten: I really like that you say, hey, start with the business outcomes first. Right. Because you know, like you could say, well, let's fix our data problems first. Well that mantra we heard like probably also five years, 20 years ago. Right. So it's very good. I think it's very healthy what you say, Tara, like start, start with the problems that you try to resolve. Right. What kind of outcomes do you want to drive and work back from there?

‍

Keith Kirkpatrick: Right.

‍

Peter van der Putten: Yeah. You know, be pragmatic a little bit about that.

Keith Kirkpatrick: Right? Yeah. Is there a situation though where if you're thinking about outcomes, is there a need to really consider overall outcomes for the entire business instead of looking at it from a siloed approach? Because I think that's where sometimes organizations can say, I want to improve one little thing here. But really when we're talking about generative AI or AI as a whole, really to really get a lot of value, it needs to be something that is applied across the whole business. Right.

‍

Tara DeZao: Yeah. I would say that there's again, different stages of maturity for that. Right. Because we have organizations that have AI governance boards where they make decisions together across functional areas. And then we have, you know, a lot of organizations working in silos still. So.

‍

Peter van der Putten: Yeah, no, and I think it also depends maybe on the actual function. Right. So let's say you're more into business operations. Right. Or customer operations, for sure. That can actually be, you know, particular use cases where. An agent that solves some particular need is fine. You know, like let's say first notice of loss at an insurance company. You know, there's lots of opportunities there to improve that particular process. And there's less of a need to say, oh, it needs to be kind of coordinated with everything in the company. But if you take another example, let's say it's more about customer engagement, one to one interactions with your customers. I don't think you should do that in some form of a siloed manner where you say, well, I'm just gonna, I'm a bank, I'm just gonna focus on credit cards. Right. There's no way, you know, like in the moment. Yeah, I think you would need to have to consider everything that's in play. Right?

‍

Tara DeZao: Absolutely. Yeah.

‍

Keith Kirkpatrick: I mean, certainly something that we've found in our research is this idea that there is sort of a, that the organizations that are successful either have a center of excellence or they're taking a very top down approach. They can do things in a coordinated manner and then perhaps build on success in one area and kind of try to replicate at the very least the process or the iterative process going forward. But you know, I'd like to move slightly over to, you know, kind of what is PEGA doing to really help organizations go from that sort of, hey, I have this idea that I want to improve the outcomes here, there, wherever it might be and actually getting on that road?

‍

Tara DeZao: I would say that our blueprint tools are probably, you know, the way to really understand how you can transform before you do it.

‍

Keith Kirkpatrick: Right.

‍

Tara DeZao: So we have a customer engagement blueprint and we have a platform blueprint.

‍

Keith Kirkpatrick: Okay.

‍

Tara DeZao: What one is for developers and workflow automation, one is for marketers and customer engagement practitioners. And that's a way to visualize where you want to take your organization is by blueprinting your strategies and blueprinting your applications.

‍

Keith Kirkpatrick: Right. I mean, I got to play around with that. Last year when it was released and then looking at it again this year, it sounds like there's also some really interesting new enhancements there as well.

‍

Tara DeZao: There is, yeah. And it's one of those tools that we're constantly iterating on. So you know you might have a new thing two weeks after you've logged in and it looks different than the weeks previous. So it's something that has a lot of development behind it.

‍

Peter van der Putten: Yeah, like say I can talk maybe more about the workflow blueprint and you can maybe talk more about one to one for sure. So on the, I think indeed on the, on the workflow side of the house and yes we're releasing every week habit, I think we are more and more focusing also on what are the different types of inputs that you can put into that process. It all started with that you would just type, you know, like a little prompt where you say I want to develop an application for first notice of loss. Right. But where we're extending it is that you. There might be situations, especially in legacy transformation where there's a lot more background context and maybe there are some mainframe applications that people are already using today. Right. So you could have the training materials that you give to a new agent and you give those training materials as input or maybe a little video of a walkthrough demo of that legacy application. Or maybe you've taken tools from the tools from AWS or from Google that can automatically kind of extract functional descriptions from code. So there's a lot more than a richer multimodal set of inputs that you can use as input into your blueprinting process. Like how should I then re envision those processes in a modern app like Pega that allows you to also not just use it to ideate particular new use cases, but really take into account a lot of background context which is particularly important for legacy transformation. Right. And then I think yeah, on the one to one side we have some very cool new developers around multi agent type use of well, multiple agents that work together to create a blueprint. But Tara is much better positioned than I am.

‍

Tara DeZao: I mean it's really similar in terms of the inputs. So you know we started out with a blueprint being able to take your brand identity from your website and now essentially you'll be able to input your brand guidelines, potentially creative brief data sets and really activate what your customer journeys could look like. And then I think the end state down the roadmap is being able to extract insights out of the Customer decision hub and back into the blueprint. That would be the end state that we would want to, you know, develop.

‍

Keith Kirkpatrick: So it's almost like a cyclical thing that's ongoing.

‍

Tara DeZao: Yes.

‍

Keith Kirkpatrick: To constantly improve. A continuous improvement.

‍

Tara DeZao: 100%. Yes.

‍

Peter van der Putten: But with different kinds of design agents that work together. Like maybe there's a marketing analyst agent that takes that context and information from your, you know, your existing running system to see, okay, where are we now? You know, and then there's maybe a strategy agent that determines like, hey, what is the key problem that we need to focus on? And that's being used combined in both as a brief to a bunch of creative agents, you know, that is happy. You know, creative creatures create all kinds of nice outputs. And then of course you have the brand police that goes like, well, hang on, you know, like, these are nice outputs. But here you need to focus more on the, you know, the right tone of voice. For example, that our, you know, our brand personality or maybe even a legal compliance agent that says, like, hey, you know, it's great that you're offering a 10% discount, but that's not, it's not going to happen.

‍

Keith Kirkpatrick: Well, that's, you know, you bring up a really important part, which is one of the reasons why I think there's been some reluctance to utilize generative AI in large organizations is because they are afraid about things like hallucination and these things going off the rails. But what it sounds like you're able to do with this with blueprint is make sure that you can utilize the power and the creativity of say, generative AI, but you're still doing that within a framework of, you know, established workflows and making sure that you don't have something that's out of policy or completely not an output that you would be happy with as an organization.

‍

Tara DeZao: Well, you know, you do have that control. There is human control. So if, for example, creatives came back and we heard Rob Walker today on the main stage talking about something having an extra finger. If something comes back with an extra finger, the human can say, okay, this is not what we're looking for. You know, please fix it.

‍

Keith Kirkpatrick: Right.

‍

Peter van der Putten: Yeah. And I think it's in that sense a lot of organizations are talking about like, oh, we'll have these agents at runtime, but don't forget there's like a big opportunity to use these agents at design time, in this case to design your workflows or to design your customer engagement strategies, because then, then you have the human in the loop in terms of approval etc. That said, of course there is also an opportunity to, to use these agents at runtime with that. I mean, it's like I'm in the middle of this first notice of loss, you crash your car, you're giving me a call. But there as well, we really focus on how, how can you ground these agents into an ecosystem of existing workflows, existing business rules, existing policies  that need to be applied as opposed to like, oh, we'll just leave it to the LLM, we'll just leave it to the agent. Right, so that's something that we've been focusing on forever pre agents as well. How do we deal with the workflow area? How do we deal with straight through workflow or orchestrating people doing work or orchestrating between the two of them or inserting smart decisions? We're always in this environment where yeah, you need to orchestrate across all of that and it needs to be audited and regulated and whatnot. So that happens to be the perfect ecosystem for agents to exist in because we can ground them in all these other kinds of, like I said, workflows, business rules, integrations, data source, whatever is there with proper governance on top.

‍

Keith Kirkpatrick: You know what I'm hearing with all of this with the platform is it really addresses two things that I think are top of mind for organizations. One of course is time to value. Because organizations waste so much time, you know, whether it's ideating, planning, testing, piloting, going back again, something falls off. And then of course, ultimately it's the ROI puzzle. You know, it sounds like that is, you know, underlying that is, you know, two of the things that really, that the platform is able to really support is improvements on both of those.

‍

Tara DeZao: Yeah, absolutely. At least for Customer Decision Hub you can reach time to value in six months, which is very quick. And you know, that includes generating revenue but also saving on operations.

‍

Peter van der Putten: And outcomes are not also something fluffy at a high level. You know, it's really at the level of an individual interaction. I open up my mobile app right now and then recommendations are being made and I can actually see, did the customer actually like that recommendation? Did the customer click on it, and accept what not. And that's in the marketing space, but also in the workflow space because we orchestrate that entire process. We also know how long it takes and not just for the entire process, but every stage and step. We also know what was the final outcome, but also the intermediate outcomes during that process. Right. So this outcome orientation is not just something macro and high level. We do that as part of every single interaction, as part of every single process execution. We're able to track these outcomes. And that also means that then on the fly you can steer towards doing the right thing and optimize towards those outcomes. So that brings it back to this idea of the autonomous enterprise.

‍

Keith Kirkpatrick: Well, I really appreciate both of your thoughts on what we've heard here at Pegaworld 2025. I think I did this last year. I just want to ask where you think we're going to go between now and next June? And I'll throw it out to both of you.

‍

Tara DeZao: Yeah, I mean for marketers, I think we're going to start seeing, you know, we've been promising them to simplify Martech stocks for years and years and I think we might actually start to see that because for the first year ever, you know, I track the state of Martech through different reports and such. We didn't have double digit vendor growth this year for the first time in about 10 years. So I think we're going to start seeing some market consolidation.

‍

Keith Kirkpatrick: Okay, interesting.

‍

Peter van der Putten: Yeah, I think any AI space, AI and workflow space, we'll actually continue to see consolidation as well, funnily enough in more in the underlying gen AI services. Right. So better, faster, cheaper gen AI models. I think don't underestimate Google, you know that we've seen that like it's no longer that OpenAI, it's just, you know, the single best solution on the market and likewise from AWS or other vendors. So I think that will continue to commoditize that market through this hyper competitive nature, which is good. Which actually also then implies that most of the values in the application layer on top. Right. So people like Andrew Ng have said the same thing. The application layer where AI and automation comes together, that's the sweet spot, you know, guess where we are exactly there. Right. So your question was, what's going to happen? I think we'll see that commoditization at the bottom. I think we'll see more maturity in that application layer because people were first focusing on the cool kits, you know, like just having some agent toolkit and building something from scratch but these enterprises that are here, big banks, telcos, governments, organizations, they're not greenfield organizations. Right. So they're looking more for in this example than agentic systems that are embedded into an ecosystem with all these regulations and workflows. That is already out there. Right. So I think that's, that's also a sign of maturity at a more mature enterprise use of agents. And I expect that next year in this conference we'll have at least half of the breakouts, people presenting real agentic use cases that they brought to life. All right.

‍

Keith Kirkpatrick: All right, well, then we'll leave it there and we'll check back in next year at PegaWorld.

‍

Tara DeZao: Thank you.

‍

Peter van der Putten: Awesome. Thanks a lot. Thank you.

‍

Keith Kirkpatrick: All right, well, thank you all for joining us. I've been Keith Kirkpatrick at the Futurum Group, and we'll see you all next year.

‍