From COBOL to the Cloud

Hi everyone. Thanks so much for joining today. Welcome to this timely and exciting webinar on AI powered legacy transformation. What does that mean? How do you get started? What else do you need to know? We're going to cover all of that and more shortly here. I'm Sean Callahan. I'm a product marketer here at Pega.

I'm coming to you from our studio in Waltham, Massachusetts at Pega HQ that's just west of Boston for those keeping track at home. And I'm thrilled to introduce a couple of very knowledgeable guests on this topic. First up, I'll introduce Bruno Sahinoglu from AWS. He works on the global team dedicated to mainframe modernization. Bruno, thanks so much for joining today. Thanks. Thanks for having us. And I'll also introduce from Forrester Research, Bill Martorelli. He's a principal analyst.

Bill is the author of the Forrester Wave for Application Modernization and Migration Services. If I have that correct. That is correct. Thanks so much, Bill, for making it. Hey, it's great to be here. Thanks for having me. Um, so we'll get started here in just a second and I'll start diving into this topic. But I also would like to give a brief introduction to Pega. For those of you who aren't as familiar with what we do.

I think Forrester and AWS are pretty common names. Pega maybe not so much. So if you're coming at us from a non pega audience. We are an enterprise platform for AI powered decisioning and workflow automation. And we work with some of the largest, most complex enterprises in the world to re-imagine outdated processes, manual processes, manual work. As modern AI powered automation imbued workflow applications that are built for the cloud, built for an AI powered platform built for SaaS, built for the future. And so obviously, from that standpoint, we're very excited about what's to come with legacy transformation and all of the new technology and capabilities that are available now. And our clients are clamoring for this. Everybody is looking for ways to transform faster and more effectively than what was previously possible.

And so with that in mind, I think we can kind of dive in. And so, Bill, I'd like to start with you. You were, like I said, the author of the The wave for Application Modernization, Migration Services. I know the last one came out, uh, at the beginning of 2025 and March, right? That's right. And, um, I know in that you mentioned AI as being a looming technology, something that's, you know, definitely coming shortly, but maybe hadn't quite had the impact in that moment as it maybe has had since, uh, what's animating you as you think about this industry now with AI? And what are your clients asking you about what's changed since then and now and maybe in the future? Well, it's interesting you say that because when we started off on our report, which was probably in the latter part of last year, uh, we, we did find there was a tremendous amount of excitement built up for generative AI. And little did we know that by the time we moved to publication, the whole world kind of shifted under our feet and Arjuntech really popped.

And so a whole nother set of expectations, a whole nother set of possibilities, you know, really, really emerged to the fore. And I think that, you know, I think it was somewhat of a double whammy in the sense that, you know, customers really hadn't really known or come to know exactly what generative AI could do. And then, of course, the agentic foot, the other foot falls, so to speak. And, uh, I know that the, you know, the clients, uh, that we have are very, you know, excited about the possibilities. They're certainly very mindful that as many have told them, right? If you don't engage in this technology, your competitors will. But yet at the same time, there's a tangible hesitation as well that that, you know, at least respects the risks, respects the, the, the challenges presented by these new technologies. So it's a very interesting time to, to really be talking about this. Yeah, I want to pick up on some of the risks and misgivings in a second year.

But, Bruno, I'll turn to you. Basically the same question. AWS, you guys have, you know, cloud migration and modernization has sort of been the name of the game for you for years and years. What are you most excited about right now? What's what's AWS? What do you have your mind on with with AI and with this new sort of approach to modernization? That's true that AI has been a game changer, and we have been able to witness this from the firsthand. When we started talking about mainframe modernization, if this is a topic today, but it's work for any kind of legacy transformation. We realized that there is no one fits all solution for this kind of workload when you're migrating, because it's not about moving from one language to another.

It's about governance, transformation, testing, performance. A lot of things. It's a whole ecosystem. So we started to build a partnership with partners who have delivered this kind of project on our platform. That was our competency. It was like back in 2021, there was no AI at this time. It was traditional approach. If you heard about refactoring replatforming, that was the way to go. The less risky approach for transformation and migration.

And then we started to work with ISV as well, having acquisition and also strategic agreement with some of those to make those tooling more affordable and more accessible for partner and customer. And in 2024, when we also embraced the AI technology to try to accelerate even faster to this kind of modernization. We announced the GA of AWS transform with multiple, uh, version and one specific for mainframe. Um, we also set up the same year a partner roundtable to see how does partner who was building their own methodology, bringing, building their own tooling. All together, it was called Genesis toolbox in 2024. But, uh, during the next year, in 2025, we started to work with Pega, as you know, and some of the partners to try to take this idea into a solution to bring all of this methodology, tooling, knowledge industry all together into what we have just announced in Reinvent 25, which is the composability framework around transform. And that's where we see AI playing a new role where we had traditional approach back in time. And now we see new player who can tackle these kind of complex migration. Yeah, I think that's the ecosystem is a great point.

And we'll definitely talk a little bit more about that later. I think that's all fascinating stuff. And you mentioned AWS transform, which is, you know, a technology that transforms legacy code from mainframe systems and COBOL and things like that into modern, more useful formats, languages, or PDF outputs for requirements documents. I think that's a good way to make some of this stuff real for our audience, because we talk about the big picture, we talk about AI, the landscape. But the reality is, is all of this will at some point manifest as tooling and capabilities on the ground. And I'll also just take this opportunity to mention that Pega has our own sort of agentic tool called blueprint, which integrates directly with AWS transform so that a client could start and transform, refactor code from COBOL into requirements document or to other kinds of code, take that output and pull it into Pega Blueprint, which is an agentic workflow design tool that takes that sort of output along with demo videos, screenshots, BPM diagrams, other kinds of documentation, and can re-imagine those outputs as modern workflow applications that then you could take into Pega platform and on the cloud. So you're seeing AI kind of play a role at the requirements gathering phase, you know, transforming old code into modern documentation at the reimagining phase, which is taking those outputs and saying, what is this application trying to accomplish? How do we do that in a more modern way? And then once you're on a platform, you can use AI to execute the work in new ways.

That makes it a little more dynamic than it was before. So kind of the whole end to end process is now, I think, starting to be colored by some of these AI capabilities. But Bill, I want to pick up on something you said earlier, which is that there are some some risks or at least there's a, there's a, a belief that AI can, can sort of get out of control. And I would say certainly depending on how you deploy AI, it can be chaotic. We have an, a point of view on that that makes it governable and scalable. What are you hearing from some of your clients about things that they're concerned about, or what might give them pause as they consider how to deploy AI safely and securely? Well, you know, it's interesting because back in the day when we were talking about our report, the one that you mentioned and that was there were there were a lot of concerns about things like potential misuse of intellectual property, which I know still remains kind of an issue in the background. Also the idea, of course, hallucinations, hallucinations, which have not entirely gone away, although, you know, certainly we've made a lot of progress with that. I think it kind of raises the stakes to, you know, since Argentina is so much more, uh, let's say, you know, it has agency, right?

It brings agency to the equation and therefore, uh, the potential for perhaps new kinds of misbehavior. Uh, but I think, you know, I would also argue that that you mentioned this the, the question of the ecosystem, right? And that is, you know, what do you build upon now? Because I mean, it's like, there's a lot of capabilities that are emerging from various quarters, right? Every major ISV, every major hyperscaler, of course. And so, you know, and a bunch of other types of players, right, which are nontraditional players like the Nvidia's and people like that. So it's almost like, you know, what are you what? You know, I think it's important for companies to understand, you know, like the choices they're going to have to make. And that's an observation we got from your company, Bruno.

And that is there's so many players now in the Solutioning Game, right? Like seven, eight, nine different players, right? Including OEMs, you know, vars, you name it. And of course, data players. So it's become quite complex. And I'm not sure that enterprise strategies for procurement and governance are at pace yet. Yeah, I think that's absolutely fair to say. I mean, from our standpoint, you know, where you deploy it, how you deploy it matters. We like to tie our APIs into workflow, which means that you're not just telling an agent to go accomplish something.

You're telling an agent, here is the way that we get work done, and here are the ways that you can now execute that work in specific and, you know, purposeful steps. So it's a little bit more tied to a map of how work happens in your organization, which in our opinion makes it a little bit more predictable. Bruno, is there anything you'd add to that about the, you know, um, the, uh, the, the method of deploying AI securely, safely, um, governance, scalability from AWS standpoint. I would say to use it in a meaningful way would be more interesting. I had a lot of customers saying, hey, you have transformed. Yes. And you are using AI everywhere. Also to convert the code to Java, I say no, and I've been surprised. Why not using AI for this?

No. There is some space where AI do a very wonderful job and some space where there is more deterministic approach, which makes more sense. So just make sure that the customer knows when to use AI. It's the most important point I think, and not to be overwhelmed with AI will revolutionize the world. So use it carefully. Have a goal and then AI is good on delivering the goal. Yeah, yeah. Awesome. Um, well, kind of picking up in that same vein of, uh, you know, deploying securely and safely.

Um, we know that legacy transformation projects historically have tended to stall out, tend to fail, fail can mean different things, I guess stalling, going over budget, you know, ending up in a space that you didn't expect to be in at the output of that. Outside of that project, I think we have the stats. 70% of enterprise transformation projects fail, so there's probably some reasonable risk aversion for some of these projects. Bill, as, as your clients and as our clients consider moving off of mainframe systems or outdated systems into cloud based systems. Um, what gives you encouragement that now is a better time to take some of those projects on? Maybe there's an IT leader who's been burned by a project that went over budget that took too long. Why should that person now say, no, no, no, there's reason to look at this again with fresh eyes. Well, you know, it's I think you're right when you say that, that that, you know, the way people interpret failure, it does vary quite a bit. And we're always wondering like, why or how did we get to the point where we might even call it failing, right?

And depending on, but I think, I think you're right to point that out. And that is. Yeah, I am encouraged by the fact that There is a. And I've seen this in the work we did on the report. You mentioned the wave that companies were very concerned about the the business justification for what they were doing, both from the migration standpoint as well as the modernization standpoint. And I know we tend to think of migration as a one and done right where the modernization continues, right? Even what we do in modernization now, like new versions of Java, maybe, quote unquote, tomorrow's tomorrow's legacy. But we have seen this idea that, look, there are deals out there. We see them large transactions with big service providers aimed at large scale transformation.

But we have seen a definite preference instead for incremental types of projects, right, that emphasize near-term results. Uh, even to the extent that they're self-funding in the sense that they, they emphasize initially a pool of cost savings that can be applied to the modernization effort and on an ongoing basis. So I think that, that, that mentality, you know, and it's kind of implicit in things like the strangler pattern, right? You know, you don't necessarily do it in a big bang way. I think that's really caught on. And I guess that would give me encouragement. Yes. Yeah. That's great.

Bruno. So I'll turn to you. So what is the opportunity from the AWS standpoint for what's on the table for clients that are maybe thinking about not modernizing or thinking about modernizing in a, you know, an outdated legacy way? What are you advising your clients on? What do you think is at stake for them? So first of all, we see that AI has changed fundamentally the economics and risk profile on these projects. With AI, you can start decomposing, you can assess, you can plan. It helps you to shrink those phase from six of 12 months to 6 to 8 weeks. So it's, it's very exponential.

And then with AI, you mentioned it before, we can do the cut documentation business with extraction. What was human cost task heavily back in before. And because you can decompose, you can break down your monolithic. Your monolithic mainframe. So when you just mention it, it was about big bang approach on this project. It was a way to do it and it was working. But customers have to be able to make sure that he can move everything in one time. Now, by being able to breaking down those monolithic into business function or business domain, depending on the granularity you are looking for, you can decide what you want to do. You can do augmentation.

You can just move the data first. You can build API. You can use a strangler fig pattern as a way to move and to have a no risk approach on your project. So this is where AI is playing a role on enabling those customer who was really reluctant to do the first step to now engaging with us with any other SI and ISV partner? Yeah, I think that that resonates a ton because I know when we look at mainframe projects, for example, the mainframe is this massive monolithic system, and there will be applications in the mainframe that you need to modernize, and there will be things that the mainframe does, workloads that it has that you might keep there for now. That might be too risky to move or touch. And so taking this kind of like scalpel with AI to something like a mainframe allows you to pull out the low hanging fruit and allows you also to get at that hard coded logic and rules and application data that might have been impossible to really get at previously. So you're opening up this whole new realm of white space for things that you can feasibly transform without, like breaking everything around it. Um, we've, we've done some research that said that there's like a $370 million overhead cost for maintaining enterprise systems annually in, I think in the US.

So there's a massive, um, a massive, uh, dollar amount that applies to all of this as well. Bill, did you have something to add? You know, I just wanted to make a comment based on what you said, Bruno, about the economics. And it's a very interesting question because, you know two things. One, I know that companies are struggling with the question of, of effort estimation, right? Cost estimation. And I think it's true that for the most part, the classic models, right? Function points, Cocomo what have you. Have not yet really embraced or at least, you know, fully adopted the realities or the promise of AI.

I know some clients of ours have been quite interested in that topic. Another one is, you know, the service companies are obviously very active in this whole world. And, and the idea that, you know, like the promise that, all right, if I use AI on maintenance or new development, I can gain a 30% benefit over two years or three years. And, you know, it's a very real possibility. But on the other hand, it's a real big difference saying that it might happen, uh, versus contractually guaranteeing that it will happen. And people are being pushed to do that. And it's a, it's a very interesting phenomenon. Yeah, absolutely. So we talked a little bit about the, the overall landscape of AI and transformation.

We talked about some examples of technologies. We talked a little bit about the the opportunities here and some of the costs for not transforming. Let's, let's get into maybe a little bit of the nuts and bolts and Bruno, I'll put you on the spot first and talk a little bit about how AWS is at the center of this ecosystem, which you mentioned earlier. Um, there's all kinds of overlapping tools and technologies you guys play a part in most of all of it, I would say, how do you approach, let's say an IT leaders coming to you, uh, looking for sort of the first steps of getting started? What do you recommend that? How do you recommend they start, given that there's, you know, all these different considerations to take into account technologies, all this mess of stuff out there? Um, what are some really strong basic applications, things that they can do tomorrow? So that's a good question. And to be honest with you, we don't talk about technology when we met this IT person because this is about a business transformation.

So of course when you are moving a VMware to the cloud, it's like a lift and shift on mainframe on legacy. You cannot move it like this. You have to move the application and the data. So when you open the discussion, those customers, they are also thinking, should I just move the application as it is? Should I transform the application? Do I have new business needs that I'm not able to do today on my legacy? Because many reasons. Maybe the code is trapped into the mainframe. A change in two months or three to make it happen.

I've got people maintaining the application for 40 years, and I don't have new skills to deliver something new at the cloud speed. So there are many reasons. So what we do first is we have like a discovery questionnaire. We try to understand the business goals of this modernization. And for sure we're going to gather also some technical point to make sure that we have the right solution to make it happen. So depending on the goal, we move to the next phase. And so this phase we call it the assess phase. If you are familiar with AWS migration and modernization program called Map, this is something we build over thousands of migration. So we do this assess phase of discovery.

Then we move to the mobile phase. What most of the people we we call the planning phase, where basically you try to build the project for, for this customer. And this is where we, we can use technology like ours, like transform to start the analysis to make sure that you have all the code to make it happen, because that's what's happened on mainframe and legacy is this application has been running for 50 years. Sometimes you are missing, some source code are missing and it can be a trap in your, in your project. So you can leverage some technology to make it up. And during this mobilize phase, yeah, you try to apply the right approach. So I just talk if it is a business driven project or more an I.t driven project, this is what I call the top down or bottom up approach. And it's going to depend on the customer needs here. So once we are set on the way, then you can apply any pattern of migration and we can help.

We have our tooling. We have a partner tooling and we try to apply the right tool for the right needs and we move to the migrate phase. This is the last step of the migration where most of the time on those projects so far, we had a consulting company helping on delivering this project. What is changing right now with AI are those tools are simple to use that some customers are now building their own practice to try to do the migration by themselves. That's maybe where we are starting to see some difference with the previous years. MM. That's fascinating. That's super helpful. I think I can kind of echo some of that, which is that, um, you know, the low hanging fruit is maybe the best place to start.

And I think a lot of these organizations have an idea of some of the applications that they would like to modernize and they maybe have a list somewhere of, you know, where they would get started. And I think the first couple of steps are very difficult, but it's important to, you know, take a consultative approach. I think from our standpoint as a vendor and from from your standpoint as a hyperscaler. You know, we don't expect that this is going to be like we've talked about just one giant project that's going to be contracted for the next ten years and that you're going to. Right? It's going to be let's take a look at some of the highest priorities that maybe we can help identify as being the simplest ones. And we'll start there and sort of prove out that approach and then maybe take it, take it from there in a more incremental way. Absolutely. And Bill, I know you, um, you probably counsel clients of yours fairly regularly about where they can get started.

Sure. And I'm curious from your perspective too, um, you know, this is also like, there's a human element to this, right? And probably a lot of the big challenges with starting out on a project like this come to do with change management and mentalities and things like that. What do you, what do you talk to your clients about when it comes to starting? Starting out a project like this. You know, it's interesting you say this because on the one hand, you know, some companies, it would be wise for them to start small and get some quick wins, right? But on the other hand, we see some really aggressive clients that almost take the reverse approach. I'm sure you've seen this where they like to like do the most, the most difficult thing first so that they can prove, you know, the concept and then move on. And, you know, I think that makes a lot of sense.

One of the other things we, we hear, and I've heard this from clients tell me this, and that is we've argued. I know a lot of people have argued that that you that AI is like a new user interface, but modernization, you know, at the end of the day, it's got to be more than skin deep, right? At least at some point. So, I mean, you know, we've, we've heard clients struggle with this internally to, to convince colleagues that, look, you know, we need to think about the back end too, right? The system of action as we're hearing these days. And, and so I think that, you know, if I were to say, where do you begin? I mean, you know, I think that that it depends on a lot of factors, including the urgency with which the the modernization imperative is being faced. I mean, some companies face, you know, virtual extinction if they can't deliver a really powerful customer experience, right. And their competitors do.

Other companies, not so much. Right. And so then for them, I think, you know, the stakes are not the same. And I think that may figure into the decision making as well. And just to emphasize what he said. So we have two kinds of customer. We have customer who said, hey, I have a batch on my mainframe running every night, but now I'm moving to the cloud. I want more near real time, event driven workload where my customer can have a new experience on the same application but have up to date data every, every second. We have other customers who say, I've got my renewal contract coming soon, or it can be a license or the hardware.

It can be any other reason, and they want to move to the cloud as fast as possible. And depending on the needs, you're going to have to propose a different kind of approach and solution. Yeah, absolutely. And Bill, I think that was a great point you made too, where it's not necessarily just taking the easiest application. You want to take something that does, you know, it does take a little bit of effort to show that the technique will actually work, and then that can subsequent projects become even easier because you've done something that's got a little more meat to it up front. Yeah. And I think orchestrating that, the difficult, the easy can help maintain a sense of commitment, right? When otherwise enthusiasm, you know, might be lagging, right? Like we talked about the large scale transformation projects that that fail.

Sure. Yeah. Right. Absolutely. Um, what I'm going to put you on the spot, bill. We actually we did not prepare for this. So if we want to edit this out, we can, but just I'm curious, what kinds of systems are you encountering? On the most part, we, I mean, I know mainframe is a big focus for, for Bruno, it's a big focus of ours, although we also encounter like Lotus Notes and some of these BPM platforms. Is that kind of in line with what you're seeing out in the field regularly?

Yeah, I mean, I would say so a lot of the, the, the, the clients are talking about, you know, some sometimes they think in terms of moving, let's say their, their key, you know, customer facing kind of systems to a more modernized, uh, architecture and presentation layer. Uh, sometimes they still tend to think of it in terms of on a percentage basis, right? In terms of, uh, and that, of course, you know, may, may enter into the thinking when an impending event like a data center closure, co-location contract, ending whatever is making people think broadly, right? And so, uh, I don't know, there's a lot of different things going on, but I think that, you know, with Agentic, obviously we're tending to see a lot of use cases now focusing on business processes, right? Uh, uh, and, and other, and I think there's also, though a corollary of this, that that will occur internally, right internally to it with, with cloud management, with software development and all those kind of things. So it's difficult to generalize, but there are a lot of, uh, patterns, let's say, that are, that are forming. Yeah. Great. Um, let's talk a little bit about if we can, some, some stories because, you know, we've been talking a lot in the realm of, uh, theory, I suppose, or anecdotes.

Um, I don't know if you guys have some examples you'd like to share. I know from our standpoint, we had a client, um, recently speak at reinvent about a project that they're undertaking with, with us and AWS. Um, and they are, I'm going to name names here now because they spoke at reinvent. This is Unum. They are a large insurance company and they were looking at modernizing, uh, some of their applications and had had gotten, uh, massive, you know, quotes from other vendors about long, long timelines that you wouldn't believe in a huge astronomical number for something that was a bit more of an all in approach. And they came to us and asked if we could do something a little differently. And so we are taking this approach that I mentioned earlier from AWS transform into blueprint and working with them to target some of their most critical applications, that they're desperately needing to modernize fast and prove that as a proof of concept. Um, and it's early returns. Again, it's a proof of concept.

Maybe by the time you watch this webinar, they will have spoken elsewhere and there'll be more details. So please, uh, you know, go to pega.com and you can find their story at some point here, hopefully. Um, but in the early returns are very promising. And just showing that this is a technique that does work. Um, Bruno, are you seeing anything else from your standpoint that you want to mention here? So maybe on unum just to, to, to come back on this example, what is interesting and fascinating actually is mainframe, who has the mainframe today is any company who have more than 50 years old, basically. And I think Unum is like born in a, in the 1847 I remember well. And the claim processing they tried to modernize here is because they was using probably punch card before having computer because mainframe born in the 50s. So those processing Was, uh, in multiple and aspect.

It was not only on mainframe was multiple platform. And when they come to you and say, I want to modernize my workload and, and give basically a new experience to my customers. So they have a very good goals that we try to say how we can leverage AI to help this customer to give this new experience to their customer and to simplify the entire claim processing journey. And we were talking about 1.5 million line of code of COBOL plus other many other processing outside. But we, I think they just mentioned we saved them 7000 hours of manual processing on using our solution plus yours together to be able to build this new end to end processing, extract the business rules and put everything together, validate. And the funny thing is we did it without a lot. Maybe there was one COBOL developer on our call. So this is where AI and hygienic and all of this solution together is very exciting for me is we don't need COBOL developer or I say cobalt, but cobalt. We can still find a lot of COBOL.

The other, the other niche technology and languages on mainframe where there is no more people able to understand those applications. So this is where I think I'm very excited about the future. Yeah. That's awesome. And it does. It does give you a good I mean, for a company that started to make me do the math 200 years ago, 150 years ago or something, it is kind of, you know, you realize that you start building processes on the technology that you have at the time. And it's hard once you start building this out more and more and more to then go back and say, well, this, this one that's four layers deep is, you know, something that we should try to modernize when it's just captured under, you know, so much, so many layers of systems that were new at some time but have since become outdated. Uh, so, you know, I have empathy for companies in that situation, especially when you've looked at, you know, transformation timelines that you say, okay, well, we're going to sign a contract that's going to be a decade long. I don't ten years.

You know, I mean, what was happening ten years ago, I don't even remember ten years from now. How are you going to make any guarantee of the of what the technology landscape is going to be like? I don't know if I have the opportunity to give you another example, please. It's in Latam. It's also one of our public case studies, Itau Bank. So they started a project of modernizing their legacy. And maybe we have to define what we mean by legacy. Here we are talking about mainframe. But there are so many different kind of mainframe.

But in this specific case, they started ten years ago. So there was no not such of AI at this time. But they have the plan of modernizing their mainframe from a business perspective, not from an IT perspective or just moving the cloud to a new platform. No, they was looking, hey, we want to help our customer to have a new, uh, solution in their hands. And because at this time, there was also new fintech company born on the cloud was starting to compete with traditional banks, especially in in Brazil. The bank was called Nubank and Nubank is also born by coincidence on AWS. And like in few years, they got like 100 million customers and they say what we can do. So they started the modernization. And with AI, they just announced that they already migrated like almost 60% of their mainframe, and they planned for 2028 to shut down the mainframe.

And we are talking about the bank. So we are not talking about 1.5 million lines of code. We're talking about 100 and 100 million line of code, and they are able to do it. They are on the right path. And if we look at the market, I did like a market analysis and they have been able to catch up with the fintech. And when we compare with traditional bank in in the Latam region where they are still doing modernization on prem, they are not going as speed as Itau Bank. So it's pretty impressive to see how AI is impacting a real world example. Yeah, that's the competitive, um, motivation is powerful and it seems like it's AI is kind of leveling the playing field a little bit. Bill, is there anything you want?

I think there's a lot of stories like that. And I think, you know, the fact that there are so many opportunities does belie that question of maybe there are those savings that are available. I'm not arguing that people should go out and just demand, you know, 30% off the top. Just, you know, but but I when you think about that, you really do see that there is a huge opportunity out there for, for, for, for savings, for transformation, uh, that, that AI helps to make possible. Yeah, absolutely. Um, all right, well, let's pivot now a little bit and look at the future. So, um, I guess, um, let's start maybe with you, Bruno, because we just kind of came from a place of the ecosystem and we talked a little bit about customers. What are you most excited about? Um, looking ahead to what's coming down the pike for AWS in terms of modernization.

What, um, what are you looking forward to the most? And I'll make the caveat here that this webinar will air for a year. So in a year from now, things may be different, but let's just work from what we know from today. I mean, when we had this discussion in 2024 talking about how to integrate partner into, into our solution, we didn't know how we're going to make it happen. And then suddenly we have MCP server coming from nowhere and hey, let's use it. So for sure, maybe something will change while we are recording this podcast. I think first it's the AI will help on the spec driven. So, so far when we talk about traditional approach in mainframe modernization, it was using reading the COBOL, using solution to read the COBOL and create new languages, or taking the COBOL and recompile it. That was a two way.

Now with AI y, we can extract business rules and documentation. We can use this information to generate specs. Basically, switch DOS, COBOL languages and documentation into English where a user can understand what was the purpose of this line of code? And with English natural languages, we can use AI to generate new applications in the way you want. So new developer will not have to learn to code. They will have to learn to generate good spectra. That's that's that's, I think the future. The second one is the low code. No code like we are doing with Pega.

So when you try to target a specific application domain or specific domain function, you can use the spec driven approach. When you try to tackle a bigger chunk of mainframe, there is a lot of solution. And that's what you are doing with blueprint is to read this million line of code of COBOL and to rebuild from the documentation and business rules. The entire end to end process. That's the new way, I think, to modernize an application on mainframe. And the third one is I think the AI will keep evolving. And we just announced it recently is the autonomous agent. AI needs context and sometimes you struggle with the context. Now we have an independent worker that can take a task and do a long time processing.

So we are not talking about proactive, it's AI. Working with AI worker autonomously and these things for next year, that will be a game changer as well. Amazing. Yeah. Bill, what's most exciting for you as you look ahead and maybe if you would, what's, what's animating your thinking as you think about maybe another wave or how this industry might change? Well, I think that I think you raised an interesting point that combination of, of, of AI and, and no code low code does promise a alternative view. I guess you could say of of application generation. So we've, we, we think that has enormous implications. When you think about the, the, uh, the, the, the boundaries that we've gotten used to between the, the major software players like, you know, your major ERP players and the like who are all vying, of course, to be major players.

Right. And, and, uh, to some extent the, the, the, the market will kind of evolve into separate distinct layers, right? That composed of different companies, uh, working together. Right. And, uh, because it's kind of unusual or it would be much to expect that there will be simply one champion agentic player that emerges from the chaos and rules, you know, the world, right? And so then, like you mentioned, uh, your, your, uh, whatever that thing you mentioned MCP, right? Of course, you know. And other interoperability standards. And, you know, the role of orchestration is going to be a big opportunity for the for the service companies as well, which we think will be, uh, be very, very interesting.

But, you know, you raise an interesting question because every time we think about a wave, we think about what makes sense from the standpoint of the then market. And the last time we did it, we combined the application modernization with multi-cloud management services, which I think are heading into a new, a new era, right? Given the fact that the, the AI, uh, aspects are becoming and introducing greater degrees of complexity into the environment and also, you know, teaching us that, you know, the major hyperscalers are not necessarily equivalent 100% of what they're bringing to the table with regard to AI and their preference for large language models and so forth. There are partnerships with OpenAI and whatnot. So, you know, that that I think is something we'll probably split off into another separate wave as we had it in the past. We'll see. We'll see about that. Yeah. Yeah.

Excited for that. I'll just add my $0.02. I'm. I'm thrilled to see some of these agentic and autonomous systems really start to land. I mean, it feels like, you know, this is it's so prevalent now that it feels like we've been doing this forever. But it is easy to remember that, easy to forget. I should say that in the beginning of 2025, this was, as you mentioned in the last wave, like kind of, you know, looming, I think was the word you used, but but not fully there. I was at a partner conference in March of 2025, starting to introduce this idea of legacy transformation and AI transformation, and we were still figuring it out. We were still figuring out who our ecosystem players would be and what exactly the tooling would look like.

And so much has changed in that time that now I think we're ready to say, okay, you know, AI modernization is a killer use case for AI. And I think that we're going to really see that play out over the next year. And I think that's going to be a really exciting time to observe. Um, Bill, I will I will ask one more question of you, which is, um, given that rate of change, how are you helping your clients to prepare for these future unknowns? Well, just to try to. I like. The. Question, use your crystal ball here. And, uh, well.

You know, I think that, uh, the, the important thing I think is I try to allude to it earlier in the sense that try to make a coherent, uh, sense of it in the sense that, you know, we know that some of the boundaries are blurring between some of the major participants, but make the make as intelligent a decision that you can at each layer, understanding that you may have to make another decision right in the not too distant future. And, uh, in that light, I think. Yeah. And, and also, of course, you know, introduce and be mindful of the, the governance requirements, you know, the guardrails that need to be in place. And of course, the, the human factor, right? The change management, the behavioral aspects that yeah, that, that can really make or break you if you're not careful. So I guess that's how we would generally do that. I think that's a great answer. Um, at Pega we like to say build for change.

So you could, you know, take that off if you ever want to use that. That's a good viable way to keep that in mind. Yeah. Well that wraps up for today. Thanks so much for joining us. Uh, you can always learn more by going to Pega Blueprint. If you want to start playing around with our Agentic workflow design tool, that's where you can do it. Um, stay tuned for more to come from this. I'll also plug Peg world 2026.

We are airing this webinar for about a year, so it's possible that you may come to this after Pega World has already completed, but you can still go back to our website and check out all of the sessions that you have to see there where we have clients, uh, partners like AWS and others coming on stage to talk about legacy transformation and all the stuff that's coming down the pipe. So thanks so much for joining. Hope to see you soon. Bye bye.