The Future Of AI-Driven Design | with Alan Trefler

Thank you, Steph. You know, I like being called the founder and CEO because, you know, founders are forever. CEOs, they come and go, though. It's been a while.

Never.

So the last several years have been really amazing in terms of the evolution and the broadening of the availability of design thinking. I think thinking in a design context is super important. But what we did, you know, starting a little over two years ago is began putting a tremendous amount of effort in getting real results and being able to incorporate a lot of design thinking elements into this Blueprint technology, right? Which really from my point of view, it really democratizes design thinking instead of having a group of design thinking specialists who go in and, you know, bring the magic powder to bring design thinking to the folks who might not have it. This allows the combination of roles in a way that's super exciting and lets us through our Blueprint technology and through the magic of generative AI incorporate design thinking in the way that people capture what they want a system to do and how they want to run their business and even to be challenged by Blueprint to do it better than they might have imagined that they could. And to me, this ability to take a lot of the principles of design thinking and have it work in conjunction with AI makes it just way more available and tremendously more powerful.

Well, you know, I think on one hand it does and on one hand it doesn't because certainly you want to look at what you're doing from the point of view of a design perspective but many of the things that people used to agonize over. So I'll just give you an example. We would routinely when we worked with a customer go in and send a person who was like a certified design thinker, right?

Well, you... I look the part, I look as certified as I think. I am a certifier.

But the reality is what we're really trying to do is make it so... well, I'll tell you what the process used to be at least when I was involved and you can tell me how this matches, Clay, what your experience was. We would send a couple of these, you know, design wizards in to a big conference room and there would be a huge whiteboard and people would draw on the board and use stickies to debate and they would agonize, am I getting this process right? Am I just redoing what I've done historically? Am I really doing things that from a user perspective are going to be a good experience or a bad experience?

And this had a bunch of problems. First, it's very intimidating to start with a blank whiteboard or a clean sheet of paper. So, I think it kind of put people on a mental defensive to start with. Secondly, one of the principles at Pega that is really important is what we call center-out thinking, which I think represents really an important aspect of design thinking. Center-out says you want to design your system around the work your business or organization wants to do. So you don't want to design it around screens, though obviously getting the screens right is going to be super important. You want to say, 'Hey, what are we in business to do?' And what that means is everything you design should be intrinsically multi-channel or omni-channel, should be intrinsically available to agentic processing because you're really designing the work itself. And then if you go through Blueprint, it's not till after you've designed the workflows and the stages and the steps and had gotten a lot of the data right. Only then does the AI show you what this might look like in a design and professional way across the different channels you might want to run in. That's a real reversal of the traditional 'I'm going to stare at a whiteboard. I'm going to agonize over it and try to get it right.' And we find that it's completely changed our business. What's interesting, I do maybe a confession I should make here is when Blueprint came out and Pega started talking about Blueprint, I was like maybe I need to go get another job, right? Because the AI is going to come and get me, right? And in all honesty, just to step back from that really when Blueprint first came out, I was like, 'This is a great sales tool for Pega.' That was my first inclination.

Right.

Well, I think that what's so exciting to me about the way we've seen the design thinking morph, right, as a result of this technology is it really makes it so that people don't have to confront that blank page or they don't have to go through that, you know, that agonizing thing people would do where they would say, 'I'm going to define my as-is processes in agonizing detail and then I'm going to, you know, suffer through transforming them to to-be.' No, you don't have to do it. You basically just explain this is what I want my business to do. And what we're able to do is take our design patterns and we've put lots of intellectual property into the way it works. We were able to go out to the internet and say, 'Hey, is there a best practice for doing some of this?' We're able to take input from guys like you, Clay. You can take, you know, chapters of your book and feed.

Well, you could in all seriousness, you could take the sorts of things you teach and give it to them. Give it in. And it does a really interesting job of giving you a way to think about your business problem in this center-out fashion and in a way that helps bring together the elements that always made building this sort of complicated. Brings together how am I going to interface to backend systems? How am I going to work with people? How am I going to define what's going on? And, you know, it keeps getting better. Every two weeks, we put out another release of Blueprint on pega.com. And, you know, the last one that just came out, it's now coming up with the decision tables and business.

Yeah. And so as we make it more powerful, I do think that this just really empowers people to think about their business differently and actually be challenged by the technology as opposed to having to wrestle with the technology.

And I think that's right and it's a real interesting development because in one of the earlier versions of it, there was a little button that said 'preview your app' which flashes the... which actually creates the screens and the user experiences when you click it. And we had that button available pretty early in the process. And people just love to press it because humans love the screen.

And the real thing is, hey, what am I trying to achieve?

I've got customers. So look, we have had over 200,000 blueprints created. That's crazy. Lots of them are being... some customers, they wanted to know if we minded if they used it to create documentation that they would just take the PDFs that come out of it and give it to their auditors to explain what their process flows. Said, 'No, you know, my take is the more people put into Blueprint, eventually good stuff will come out because eventually you're going to want to automate some event, that sort of thing.' So, I think that's just fine. But it is trying to encapsulate a way of thinking to be experiential. That to me is the big change. A lot of the way we used to build systems and a lot of people still build systems is very theoretical.

But when you actually can experience it in the course of a pretty quick period of time and then react both as an individual or invite other people to collaborate on it, etc. You're really creating a canvas for better creativity and just a more relaxed way of going at, you know, the really hard work of change.

Well, I think that there's a fundamental question in the whole AI and particularly this agentic universe here, which is the way some people are using it is they go to some prompt studio that a company like Microsoft or ServiceNow or Salesforce has and they go design the perfect prompts. And what they then do is they run those prompts to really try to create answers in a customer service setting or to be able to have somebody sign up for a loan or be able to do a variety of things. That's a very standard way that a whole group of companies are thinking about and talking about how they want to design in this AI world. And we think it's completely wrong.

Because what they're doing is it's abdication. Now, they might say, 'I'm designing this prompt.' But that's not really a holistic way to think of and to understand your system and what its outputs will be. And let's face it, these language models, well, they still hallucinate.

To the data that's available to them at the time that they're making decisions. And there's no ability to observe the design, right? There's only an ability to observe what happened and maybe understand why it happened after it happened. That doesn't work in regulated industries or that doesn't work for things where people quite reasonably, I think, want to say, 'Hey, we're going to do design and design thinking as part of the build process and we're going to build something that we create workflows from and then at runtime when somebody actually wants to use it, we're going to use the language model just for English and just for or whatever language translation, just for semantics, which it never gets wrong.' What that lets you do is use the creativity and the power of the language model to its absolute fullness because you're doing it at design time but at runtime you've got the reliability of workflows that you can have curated, you can have observed and you can have enhanced.

Right. So that's how I think AI is actually encouraging a lot of the antithesis of good design because it's really saying, 'Hey, just turn it over, let it do it.' And, you know, look, if this AGI, you know, artificial general intelligence comes, who knows? Maybe we'll all be replaced by a prompt.

Well, Blueprint is really what some people might refer to as a vibe development environment. Blueprint.

We're playing with it. But I'll get some design feedback on it from you guys, I'm sure. But Blueprint really is a way to interact with a set of agents that are able to read documents you might put in, take flowcharts, take input, take textual input. You can actually talk to it. Be able to take all of those capabilities and propose how your business should run whatever functions you're telling it about. And then you get to interact with it and you get to, you know, vibe on as it were and be able to change it.

But yeah, no, finish your thought. But the cool thing that we think is so exciting is Blueprint transforms itself into the Pega model which allows you two months, three months later to go back and change things. As opposed to a lot of these folks who are doing their vibing but then they try to generate code and the problem with that is it's very, very difficult to go back to a corpus of code and go back and say I want to make these small changes because we got a model there. We think it's a really exciting way.

Well, I can see you're betting you're not going to be replaced with a prompt.

There are people there, you know, who are really into that. And look, prompts have their place, don't get me wrong, but if you're really trying to figure out if you're just going to like lend money to this person or that person, you better treat the customers in a way that's going to be understandable and consistent with both ethical and legal requirements. It's very difficult to do when you just turn it over to the machine and how it's been trained.

Well, you know, let's just say valuations around things that involve those magic letters AI have become a little frothy and particular.

Well, you know, I'm not going to cast aspersions on any company that's done as, you know, candidly amazing job with the technology but, you know, valuations I think are not the way to judge the ultimate success of a company. The ultimate success of a company is judged in how its customers do and are they able to use it? Is it sustainable? Is it something that's going to really make that company more successful? And that's what we focus on. I think it'll be a ways before you see the one-person unicorn, but, you know, we as you may know prefer flying horses to unicorns. And so I would say that I'm glad that we were never a unicorn. I would view that sort of an intrusion on our brand from us here.

You know, I think it can be incredibly empowering because a lot of the historical complexity can actually be facilitated by the AI. So the AI can bring to you an understanding of how to get certain things done and, you know, in addition to having people do it faster what I think is really super is it has people do it better. You know, they get better outcome as a result of.

Though you get both.

I think a lot more hands-on.

Well, I always thought that with a model-driven architecture as we improved it and made it better and safer, easier to use, we would be able to achieve, I used the word democratization before, of the role so that people who were less technical would be able to achieve things that otherwise required great technical acumen or if you had a business that was going to scale to a high level that you were protected from certain types of mistakes. I think that this has really, the AI and Blueprint has really from my point of view amazingly accelerated this democratization.