Software That Never Breaks: OutSystems CEO Woodson Martin on Building Enterprise-Grade Apps at ...

Introduction of Woodson Martin

0:00Hello, and welcome back to the Cognitive Revolution. Today my guest is Woodson Martin, CEO of OutSystems, the pioneering enterprise software development platform that's helped many of the world's largest companies build and run complex, mission-critical applications since its founding in 2001, and which now specializes in helping those same giants build, orchestrate, and govern agentic systems. Woodson took over for OutSystems' legendary founding CEO, Paolo Rosado, in 2025. So we

0:33start with his thoughts on leadership transitions in the AI era, including how he assessed the company's AI readiness and what it was seeking in a new leader. From there, we go on to discuss what makes a software platform enterprise-grade in today's world, with Woodson emphasizing the importance of human trust established over years, and also the critical role that well-designed software abstractions play in allowing coding agents to accelerate the development process while still delivering on the company's core promise of software that never breaks.

1:07For OutSystems' customers, which include Petrobras, Brazil's state-controlled energy company, Vodafone, one of the world's largest mobile carriers, and Toyota, the world's largest car maker, among many, many others, this new paradigm makes it possible to accelerate product roadmaps, to the point that some of them are actually starting to clear what had previously been ever-growing development backlogs. Woodson says that most of what their customers are shipping today is still traditional, deterministic applications that don't really require frontier AI.

1:40But internally at OutSystems, they've gone from four major feature releases in Q4, to 19 in Q1, and 26 in Q2. And meanwhile, they're rapidly agentifying their own business, with major initiatives including a customer-facing application development assistant, which they call Mentor. An in-house AI infrastructure investment that's built a new harness and a custom gateway, which has allowed them to reduce their token spend from a peak in June and July to being below their projections today. And a new agent foundry that they're using to suggest new agentic systems

2:14to build for customers and forecast the associated ROI. Toward the end, we get Woodson's strategic analysis on the future of the software industry, including how competition will shape up as platforms expand to serve many adjacent niches, what companies should do to differentiate themselves as everyone begins to do everything, and why he's super bullish on AI-pilled junior talent.

2:41From the outside, it's often said that for all the amazing things that AI can do, we still haven't seen measurable impact on productivity or global GDP growth. But I think you'll agree, after hearing Woodson's perspective from the helm of a platform that serves many of the world's most consequential and often conservative companies, that change is indeed very well underway. With that, I hope you enjoy my conversation with Woodson Martin, CEO of OutSystems.

Taking over a foundational platform

3:11Woodson Martin, CEO of OutSystems, welcome to the Cognitive Revolution. Thanks, Nathan. Exciting times. Good to be here. Yeah, boy, never a dull moment in the AI game these days. I wanted to start with kind of a sociological question for you, because in doing my homework, I learned that you took over OutSystems from a legendary founding CEO who had led the business for 24 years. And this just has me thinking, what an interesting time to go through a leadership transition at a software company,

3:44to be in your position of taking over a software company, to be a company looking for a new leader. That stuff is always fraught. But in the AI era, there's this whole other dimension of like, where is this company on their journey? Where is this leader in terms of his vision is? Or who gets it? Who doesn't get it? So I'd love to hear the story of the thought process and how you guys got to understand one another well enough to know that it would be a fit going into the AI era. Yeah, great. I'll start with Paolo Rosato, who we're talking about here. Paolo's a badass.

4:16You know, he started OutSystems with a very bold vision back, you know, in 2001. And that vision endures. And it was the idea that like custom enterprise software and people build really complex systems, custom ERPs, terminal management systems, global financial consolidation for, you know, reinsurance companies on OutSystems as a platform. But, you know, those projects at that time, custom software projects were always years late, way over budget, super unpredictable and

4:51resulted in very rigid software that, you know, when you wanted to change it, it was another whole huge project. And Paolo was like, this is crazy. There's a better way. And his vision was simple, build faster, reliably every time and evolve software at the pace of the business without breaking anything. And that's still the core of what OutSystems does today. Of course, now we build all that stuff with AI agents doing most of the work, right? But the result is that enterprise AI

5:25just works every time, all the time. So it's kind of been amazing how enduring his initial vision was and the platform fundamentals he built. And kind of that was what got me excited and inspired about OutSystems. 16 months ago when I joined was really understanding how powerful these platform primitives that the company had built that let like relatively unskilled technical users, developers,

5:59build really complex enterprise systems, change them fast without breaking stuff. That was revolutionary at the time, but it's needed even more now, right? When AI can build all this stuff for us, the pace at which we build, the sheer volume of stuff we're managing, like that never breaks feature turns out to be even more important. So that's kind of what got me inspired about OutSystems

6:33as a business and as an opportunity. And then it was just meeting customers, like the enthusiasm for the platform, because of what that made possible for so many people and so many companies, like people just love the product. And you always want to be in a situation where your customers and your users love the technology. And so that's been a big part of it for me too. And then just seeing that some of the problems the company hadn't yet solved were things I knew how to do. You know, I came here after 18 years at Salesforce.

7:08I did a lot of jobs in that business, everything from running marketing in Europe to product managing the core sales platform to running HR for a while for the company. And through all that experience, learned a ton of things about scaling businesses. And I was like, I think I know a lot of what we could do here to really turn on the afterburners and grow this thing. And so that's super exciting. It is striking how, as I was listening to you describe the original vision, I was like,

7:38that all sounds very consistent with where we are now in the AI times, but for maybe the never breaks, right? Because we're all, I would say, at least most of us are living in a world where AI is doing a ton of stuff for us. The amount of output is up tremendously. And I'd be interested in to hear some metrics around what that looks like at OutSystems. But then we also are seeing things like break pretty often. How are you managing to maintain the value of never breaks

Maintaining software reliability with AI

8:08in the vibe coding era that we're now in? Yeah, we take a pretty different approach here to AI for software development than most things on the market today. And that's because there is an intermediate layer in the OutSystems platform, an abstraction of the intent of the application that you build with an AI. You can use Cloud Code. You can use our built-in AI assistant called Mentor. You can build with Codex. You bring any coding agent you like.

8:41And rather than manipulating code in our platform, they manipulate this abstract model of what you need. And then you press a button to deterministically, or you ask Cloud to do it, you deterministically generate the code in a way that produces an asset at the end of the day that meets all of your requirements for security, you know, role-based access controls. All those things are baked

9:14in at the platform level so that every code asset created on OutSystems respects 100% of that. And by the way, automatically reuses things you already did. So I go build with AI, I pick up Cloud, and I'm say, hey, just write me an app. It'll start everything from scratch, basically. And in a complex enterprise where I've already got hardened systems that meet my compliance requirements for GDPR or HIPAA, depending on what kind of business I'm in, I don't want that stuff reinvented all the time, go

9:45through another compliance cycle, pick up what works, add the capability I'm looking for on top of it, and push that thing out to production in a way where I know that it all works the first time. And so that's really kind of the secret sauce in OutSystems. When you're building a new system on OutSystems, you're inheriting every piece of your enterprise control plane, if you will, in every code asset that gets generated. And that's really a very different story than almost any kind of other

10:19harness people are using today to build AI tech. Yeah, in general, I've seen this pattern a lot where the incumbent that has really well-developed primitives is in, initially you have the kind of, oh, these startups might come in AI native and win the market. But it seems most of the time that having these really well-developed primitives makes it easier to put an AI layer on top of that so you can get the most out of them and reconfigure, remix them as much as you might want. Those foundations matter a lot. And it's really hard

10:55to build those primitives that work at enterprise scale in the way that OutSystems is done. And it's not just that. The software is one thing. The other thing is trust. If you're a regulated enterprise operating in Europe somewhere and you have hundreds of regulators looking at your stuff, you've got contractual obligations with all kinds of customers and business partners,

11:25you've got privacy concerns everywhere you look, the trust that you have in any system is not just a pure technical evaluation of its underpinnings, right? It's actually the entire history of the delivery of systems like the one you need on that platform that you know, and there are plenty of others out there that you can ask about their ability to trust it, what works and doesn't,

11:59et cetera. Like, so the reputation that you build up over time for that kind of trust for these kind of hardened systems and regulated industries is a thing that is just very hard to get as a startup regardless of how cool your technology is. Yeah, that's really interesting. And it's maybe the answer to another question that I was going to ask you around, what does it mean to be enterprise ready in today's world? I, of course, everybody's claiming it incumbents and startups alike, but then what I

12:30thought it used to mean was like high uptime and reliability and things like that. And now I look out at the world and I see that the fastest growing companies out there, like your anthropics and open AIs don't really have a super great uptime actually by any conventional standard, right? So it seems like they have shown that if you have the spice, so to speak, enterprises are actually maybe a little more forgiving of some of these requirements that used to be like sacrosanct than it used to seem

13:01like. How do you, what's your take on that? And what do you think really is required to be enterprise

Enterprise readiness and compliance backlogs

13:06ready today? Yeah. Okay. There's a lot here. And so, and it's not simple. Like the reality is every enterprise is a little different and what they need is different and what they need for different workloads is not always the same. So sure. In a lot of regulated enterprises, people have AI helpers doing work. There's not a lot of regulated enterprises today who have unleashed autonomous agents to run mission critical work. And that's a maturity thing for the industry part of it. Yeah.

13:40Stuff like SRAS, you know, the reliability of the system and is the model always there and able to respond uptime, et cetera. But it's way more than that in most of these organizations. And, you know, I've got customers who are well down, and I think this is true, by the way, of a lot of technology providers, customers who are well down the path of an agentic system they have built and designed and tested and they're super excited about. And it's sitting in a compliance backlog today, somewhere in

14:12their organization where it's pending the approval of the use of a specific AI model under the covers for that work. And the obstacles that get in the way are things like, well, we need to understand whether all the data used to train that model was legally acquired by the person or the entity that trained that model. Like there is a set of requirements that exist in many of these organizations

14:44that make it extraordinarily challenging to trust a workload to an autonomous agent. And this is some of these are just like, all it's doing is reading data out of PDF documents and turning it into structured data. And yet there is this type of compliance concern that slows progress. So I think enterprise ready means a lot of things, right? Yes, of course, it means the systems have to work when you need them.

15:15But it also means that they need to meet a rather extensive set of requirements that are non-functional in, you know, not just like, does it do the thing when we press the button or say the words, but does it help us meet all of our regulatory requirements, contractual requirements, et cetera? So that's, there's a lot when it comes down to what enterprise ready means in the market today.

15:46And a lot of it hasn't changed with AI. You know, some of it has that whole model provenance question. That's a relatively new one, but a lot of it isn't like, how do I know exactly how every bit of PII is being managed in this complex workflow across six systems to get this thing into production? And that's a lot of where the governance comes in, where a lot in the form of the security model comes in. So how do I ensure that the model is executing with access to only the data,

16:23that's appropriate for that. It's not consuming stuff that it shouldn't. And those are all the kinds of core primitive capabilities you need in a platform to be ready for enterprise AI. When it comes to these apps that are built and in this sort of compliance backlog that you describe, and feel free to pick it apart and subdivide your answer, but does that strike you as just prudent risk management? Or is that ultimately a mistake by leadership to not expedite some of these things

16:59in the spirit of getting the transformation that I think everybody kind of agrees is inevitable underway now? It's certainly some of both. The calcification processes inside organizations is real in large enterprises and is something that organizations need to challenge and break in order to advance. But a lot of it is also very real. In a lot of places, the regulatory constraints are significant,

17:31the liability risks are big, and reputation risk is also high. And so a lot of this is prudent decision-making. And the reality is organizations are taking different risks in different parts of their business for different kind of workflows, right? There's one level of risk I'll take for managing, for example, the allocation of IT resources to people who work in the company. Banks need that. Insurance companies need that, right? Regulated enterprises of all kinds need that

18:04kind of stuff. Take more risk there than perhaps the loan origination process where they are making decisions contestable by regulators and they need to be able to prove, right, what and how the decisions were made at every step of the process because it was all auditable. And so they need to make sure that they have what they need to be able to defend themselves in the case that they ever need to for any decision they're making. So I think it's a mix in terms of appropriate prudence and also just some legacy

18:40conservatism. What are you seeing in terms of indemnification from the big model companies? Does that move the needle for people? And what about the role of insurance on some of these things? I've previously done an episode with a company called AIUC, the AI underwriting company. They're developing standards and trying to exceed the insurance market to hopefully grease the wheels on this. Are you seeing... I've talked to a couple of entrepreneurs doing work in this domain, you know, basically AI agent insurance type offerings. We haven't actually seen that uptake of that with our customers,

19:18with the workloads they're managing on our platform. So I think that is a very interesting idea. I do see that probably could be an unlock, but I think for a lot of these organizations, it's also another unproven thing and also needs some maturation for the kind of regulated industries that often operate on our platform to like jump in on that too. Hey, we'll continue our interview in a moment after a word from our sponsors. Today's episode is sponsored by Parallel, where agents find answers.

19:53Most engineers today closely follow new model releases, but don't pay nearly as much attention to their agent's most important tool, web search. If you're like me, your agents are often conducting hundreds of searches per day. And while the unit cost is small, over time, this does start to add up. Before starting to use Parallel, I calculated that my agent search bill would total roughly $500 this year. The good news is that I just had my agents conduct a systematic test, and I found that Parallel's fast mode, which costs just $1 per thousand queries,

20:27worked just as well as my previous default provider. And switching to it will save me roughly 80% of my search bill going forward. Parallel's infrastructure is enterprise-grade, and their suite of APIs offers a range of Pareto-optimal options that allow you to choose the right balance of quality, cost, and speed for your needs. So, whether you're building voice agents that need 200 millisecond latency, or long horizon agents that need deep research, adding Parallel to your agent's toolkit is a no-brainer. Get started for free at

20:59parallel.ai slash TCR. That's parallel.ai slash TCR. Today's episode is brought to you by Anthropic. By now, you know my story. Claude drafts my intro essays, and I rewrite them. Not because the drafts are bad, but so I can stand behind everything I publish. Well, I have an important update. Claude Fable 5 is the first model to have me rethinking my rule. Today, I now think co-authorship, not sole ownership, should often be the goal. Where the model

21:35excels, rewriting its work can be more about vanity or a misplaced sense of duty than integrity. I feel it most in songwriting. I'm no lyricist, but I'm good with the song concept, and Fable writes some amazing verses. I give it feedback on its misses, and I push it to aim for higher inspiration, add layers of meaning, optimize syllable density, and above all, write a hit song. These days, I get compliments on just about every song we write together. Claude is the AI for

22:08problem solvers. It's the collaborator that understands your entire workflow and thinks with you, not for you. Whether you're debugging code at midnight, building a financial model, or strategizing your next business move, Claude extends your thinking to tackle the problems that matter. For problems worth solving, get started with Claude at Claude.ai slash TCR. That's Claude.ai slash TCR. And check out Claude Pro, which includes access to all of the features mentioned in today's episode. Once more, that's Claude.ai slash TCR. One obviously huge risk that everybody is facing

Cybersecurity and speed of remediation

22:45these days is the risk of cyber attack. How are you dealing with that? Obviously, you have a code base that goes back many years. Many people are talking about, we have to move everything to rust. We got to do a ground up rewrite. We've got to be memory safe. We got to have all these things. Otherwise, we're just way too exposed. How much investment do you plan to make there? We're making a lot of investment there. And obviously, we're thinking about two things. Like we have our own platform, our own infrastructure, our own business operations, where we're thinking

23:20about this. But we also have all the thousands of systems that our customers have built on our platform. And we think about risk in that domain. And in both those cases, the first thing we think about is speed remediation, right? Like the velocity of cyber threat is increasing, the pace with which everybody needs to be able to respond to that. We need to be able to respond to patch our own systems. We need customers to be able to respond to patch their own, their systems. And there's two worlds. Let's paint. Let's paint the world where you've got a concentrated

23:51portfolio running on a platform where there's a lot of shared primitives. There is just something a lot easier about speed remediation in that context than in the alternative context where you've got hundreds of AI built systems, each running on their own frameworks and stacks and using various open source components and so forth. And you're like, okay, now there's a vulnerability. How do I even find out where I have it, much less how do I remediate? So there's

24:23just an inherent benefit in the kind of platform approach to the kind of, hey, Wild West AI builds all the things. So that's one place where we think about the benefit of a platform like ours for speed to remediation from our customers. Well-architected systems with a lot of reuse of common components means that any remediation can be tackled kind of once in a spot and have a broad impact anywhere. And so

24:55that's one thing that we think about when we think about this. Are we upping our game? Absolutely. Are we adding resources? Are we bringing the world's most powerful AIs in to try and run our own pen tests on our own things? Are we encouraging our customers to do the same thing? Are we automating a lot of that testing and the build process and the test and the deployment process inside our own platform so that customers don't have to think about that as much? Yes. All of those things are important investments for us

25:27as we think about, you know, an increasingly risky cyber environment.

25:34Do you think that will be a sort of permanent tax on the software industry that the frontier companies just collect in perpetuity because they're constantly creating new cyber risks that only they can help you solve? Or do you envision the alternative would be like, we close all the bugs, you know, and we're in a happy place in, I don't know, 18 months or something? There is no finish line on security and cyber threat, right? Like as technology evolves and gets

26:04better at all things, it also gets better at cyber attack and the world is full of legacy technology today. So the big enterprise, we have lots of customers out there who still have 60 year old cobalt systems running in corner AS 400, Lotus notes, like all of it. And there's so many of these systems, like the idea that at some point we reach a stasis and everything is perfectly, you know, secure, I think is a pipe dream. So no, I don't think it ever gets to the point where

26:41there's nothing. Now, are the model companies who are both fueling the tech that can discover vulnerability and penetrate, right, going to benefit also from the use of those models for protection, right, and white hat and everything? Yes, for sure. Like obviously that's driving a lot of token burn today. Probably will continue to. It's hard to see why it wouldn't.

Token spend optimization and model mix

27:15Yeah. How is your token spend to the degree that you want to share the details? How has it evolved over recent months? And how do you think about it? I often ask people, what's your ratio of token spend to payroll as one way of kind of getting a handle on it? I'll say, you know, maybe one thing I'll just say is kind of peak for us was kind of June, July. It's been telling off. And part of that has been like learning a lot about and frankly, focusing on it. Like last year at this time, our sole focus

27:49was like, hey, shift everything to AI, use models for everything. And that created a huge amount, a bunch of learning first, like we learned a ton. Second, a huge increase in velocity in terms of just capability delivery. So think software engineering, big part of what we do, shipping software products. We, I think we shipped four major features in Q4 last year, 19 in Q1 and 26 in Q2 this year. So like we had saw a huge surge in our, you know, and this is not minor things. These are

28:25major new capabilities that was a dramatic acceleration due to that kind of AI investment. We saw it in the token burn, but worth it for the impact. So that was great. But one of the things that we've learned is actually most of those things that we were doing, we could do just as well, a lot cheaper with a better harness. We built one, right, for our own engineering organization that is, of course, full of all of our own context and is optimized. For us, we built a gateway, right, a route, LLM router.

28:59So we're like, not all those jobs need to go to, at the time, whatever, Opus 4Aid or something, and could cut, we could reroute to lower cost models for specific jobs. So we've actually seen, we're actually burning less today than we'd forecast in Q3 because of optimizations, largely, that we've been able to do. So I think the reality for most organizations today is that they don't need frontier models for their enterprise workloads, like almost none of enterprise workloads. And I

29:29think about the real operations of a business as opposed to necessarily the creation of a new software asset. And those workloads can use models that are three years old for most of the stuff. Like when you look, it's really boring, what most of the enterprise workloads where AI can make a real advance happen today. And a lot of that stuff can be done with deterministic code cheaper than it can be with a model at all. Or two, can be done with a lower cost model, either because you're using open weight or you're using a, just an earlier version of one of the foundation models. And the,

30:09and that's, I think a key learning for us, we're seeing that in our own work. We're seeing that certainly with customers. And this is why it's important to build your workflows, right? To build your agentic systems on a platform that makes it super easy for you to swap models to get the best of what's out there. Maybe because capability improves, maybe just because you can do it cheaper with another model. And I think we're going to see without question that most enterprise workloads are, are operating at a much lower per token cost than, you know, what the frontier of the frontier

30:45is offering today. And at the frontier, people are cutting prices, right? Like, I think that's the other thing that's happening here. So there's going to be a lot of shift. Like, I don't think we were on a sustainable trajectory, uh, in the first half of this year, right? Like everybody, every CFO got the token bill in February and March. And it was like, obviously a no shit moment for the whole world. And so everybody started doing controls, whatever they were capping token spin on a per person basis, putting in a router that would allow you to mount route jobs to lower cost things, pulling back

31:20entirely from, you know, an AI first for everything, everywhere strategy. We saw that in some companies. We saw big announcements about that. I think Microsoft, meta, Uber, like all these companies who were like, wow, maybe went too far too fast, but I think you can, you can get back to the point where you're really leveraging AI and everything. You just need to do it in a way that is smart, right? And that's why you need platforms to help you do that. And that's the kind of job our systems is doing for our customers. So let's do a little double click into that. I would be very interested

31:56to know like how you advise customers in terms of what your model mix is, what you think their model mix should be. I would guess that developers when they're coding are still upgrading to Opus 5.5 immediately, but maybe not. And then I do understand how, if you're trying to convert PDFs to structured data, obviously you can do that more cheaply. Do you recommend even going as far as fine tuning for some of those use cases to really dial in performance on a cost adjusted basis?

32:27And what about Chinese models? How are you seeing people react to just the, I don't personally feel it's all that scary, but it's sometimes the scary prospect of a Chinese intelligence in their business. Yeah. I don't know if I have unique insight on the Chinese model question. I would say that we certainly have some customers who are perfectly happy to go there either by running their own, you know, running their own versions of those, distilling them or post-training tuning. And

32:57that's happening for sure. And usually driven by cost concerns. Although I'll also say that like we have a very global business, right? Half my customers are in Europe. I got tons of customers in Asia and attitudes around this question are very different in different parts of the world, right? Not everybody is like super excited to trust their stuff to U.S. companies. And when you go to a lot of places around the world. And so we definitely see a huge diversity in terms of model, models used in agentic systems that customers build on our platform. We also have done a ton of tuning

33:31on our own. So you think about, so we offer a service today to build your apps on our platform, to build your agentic systems on our platform. We have a service we call Mentor to do that. Mentor offers its own agent experience for you as a user. You can come on to our site, you can use our IDE, and you can use our mentor directly there. But you could also operate that through MCP services, using any coding agent or harness of your choice. So what we've done, of course, is we've picked

34:08and fine-tuned a bunch of models on the back end to drive the effective and efficient execution of these builder workloads for our platform. And so when you use a frontier model on top, or even an older model on top of those MCP services, you're getting the benefit of a whole bunch of optimization under the covers, and a much lighter weight job that your harness needs to,

34:44or its models need to do to give you the result that you want on our platform. And so there's frankly just a lot of answers here around optimization at various layers of the stack that come into play. And I think one of the things that we do for our customers is we simplify all that. And we're like, well, you've done the hard work to optimize and to deliver, therefore, what you need, rapidly building, maintaining, managing complex systems with AI. And we make

35:20that easy and inexpensive, but also trustworthy for you. And we're simplifying that world. So that's a big part of the value proposition of the OutSystems platform. Hey, we'll continue our interview in a moment after a word from our sponsors.

35:56It's time to innovate at the speed of AI without compromising quality or control, which is why thousands of enterprises worldwide trust OutSystems for their mission-critical apps. Teams any size can use OutSystems to build, deploy, and manage AI apps and agents quickly and cost-effectively, without compromising reliability and security. With OutSystems, you can accelerate ideas from

36:32concept to completion. It's the leading agentic systems platform that is unified, agile, and enterprise-proven, allowing you to build your agentic future with AI solutions deeply integrated into your architecture. OutSystems, all your agentic future. Learn more at Outsystems.com slash TCR. That's Outsystems.com slash TCR. The Cognitive Revolution is brought to you and powered by Tasklet. I've been running Tasklet agents for almost a year, and they've become a

37:05major part of how the show gets made. Before every recording, one agent researches the guest and sends me a suggested outline of questions, complete with links to all sources. Meanwhile, another agent scores the overwhelming number of pitches that I'm getting from PR firms. And that's really the core idea behind Tasklet. Run your business on agents by giving them real jobs to do on an ongoing basis. Setup is incredibly easy. Just describe the work in plain English. Research a new sales lead,

37:37update the CRM, and draft a personalized reply. Tasklet connects to your business tools and can run around the clock, on a schedule, or when something happens, like a new email arrives. Each agent works with your tools and business context and remembers useful information across threads. You can also share agents, connections, and reusable skills with your team, so the workflows that one person figures out become something the whole company can use. No developer needed. You decide what agents can do on their own

38:09and what needs your approval. And there are no per-seat fees. You pay for usage, not for adding teammates. Start with one recurring job, like researching the people you're meeting. And then just keep building from there. Visit Tasklet.ai and use code COGREV for $50 in free credits. That's code COGREV at Tasklet.ai.

Balancing user flexibility and security

38:34What have you learned about balancing, and I did go use Mentor on the website, so I have a sense for the experience. On the one hand, it seems like the platform is designed to support anything or almost anything, right? You have customers that want to connect to all kinds of data sources, whatever, right? There's no end to it, I'm sure. But then on the other hand, you also have this security constraint and promise. And it seems like it would be quite a challenge to allow random users to build

39:10anything while still making sure that what they build on the platform is secure to the standard that your ultimate customer at their company expects you to uphold. So how do you strike that balance? Yeah. Well, one thing is that's not a new balance for us. That's been the story about systems for 25 years. So how do you let a relatively unskilled technologist, maybe doesn't have a computer science degree, doesn't really understand software architecture, isn't a database expert, doesn't

39:46deeply understand open source libraries, like whatever. And you say, how do you let that person build mission critical systems that run a bank? So that problem is the problem that we've kind of been solving for 25 years. And now what we do is we give the same tools that you used to have access to through a visual editor, right? Hey, let me drag and drop these things around on screens and talk about add a field for what I need. Now, of course, that's all a conversation with an AI.

40:18But under the covers, it's using the same primitives to deliver the hardened asset at the end. And it takes a lot, right? Like you have to build a sophisticated model under the covers capable of serving almost an infinite variety of use cases. So if you go, if you're wandering around, if you are a, if you have a bank account, almost anywhere in Asia, the chances are pretty good that the app you use to manage your balance and pay your bills runs on out systems, right? A ton of front ends, mobile apps,

40:51consumer experiences built on this platform. If you are a oil tanker pulling into port in Rotterdam, and unloading millions of liters of petroleum into tanks at a terminal to move onto trucks and trains and other boats to move stuff around Europe and fuel the economy, the system that runs all of that

41:21is built on out systems. Incredibly complex logic for ensuring that when you pump diesel through this pump in the terminal, you're not putting it through a pipe that last had kerosene in it, right? And you're all the quality control, all the system for managing the process of cleaning and flushing those pipes and making sure that everything works in a way that doesn't spoil literally millions

41:53and millions of euros of value of product. Those things all run on out systems. So a huge variety of complex backend, sophisticated front end, to manage that diversity of capability in a single platform. Yeah, it was a lot of work. That's why it took 25 years to do it. And that's why we're confident that those fundamentals that we've built serve all those use cases and all those highly regulated environments make it easier for our customers across any industry, honestly, regulated

42:26or not, to have the trust in the platform so that they can rapidly iterate at the experience layer, if you will, and build all kinds of new things that just always work and always work together. That's the idea.

Modernizing legacy enterprise systems

42:43Things that I noticed on the ad systems website is, or maybe it was an article with you, but somebody said that ad systems customers have a limited budget and a very long backlog. Is that changing today? Is the backlog actually getting shorter now that everybody's moving so much faster? Yeah, things are changing. So many things happening in this one. I think probably the most exciting thing happening for me is that projects that everybody was always afraid to put on the backlog are like these old legacy systems, the COBOL, the AS400, the Lotus Notes and stuff. They're finally

43:20getting their day because it's now possible with AI operating and accelerating the work at every phase of a very complicated process, by the way, just even understanding what these old legacy systems do, translating those into new requirements that are modernized, where we take old laborious work, turn it into agentic work, and then the production of the new technology, the testing, the rollout. All of those things now can

43:54be accelerated dramatically with AI. And so we're seeing customers pick up the modernization, like insurance companies, picking up the modernization of archaic 60 year old case management systems and saying, we're finally ready to take on this project. Instead of planning it as a six year thing, we're now going to do it in six months. And that is like a huge change into sort of what's on the backlog. There's another thing happening, which is, of course, like, we used to

44:29need somebody's backlog to have our, like, pretty common thing in most businesses is, hey, I need to do dashboard to track this new thing. And I would ask somebody for the dashboard, and that would go on some backlog and somebody built it. Right now, you probably, in most organizations, you're like, hey, Claude, take this spreadsheet, make me a dashboard, and boom, right? Like, you've got it yourself. So much more things have become self-service. Now, one thing a lot of companies are seeing is, okay, with all that self-service, now we're losing standardization, we're losing integration, we're going to need to go do some cleanup of those

45:05things. But the good news is, a lot of the experience has been figured out. What do people really need? Well, they built it themselves, they kept talking to Claude until they got it. So now we can say, okay, now how do I just make sure that I wire the back end of those things together in a way where they use our common data definitions, where they use our common standards for rule-based access control? And that's where a platform like OutSystems makes a big difference in processing, frankly, a very different kind of backlog than what it looked like even a few months ago. So how does this translate to the success of

Translating productivity into economic growth

45:41companies and economic growth? It's trite at this point to say, like, computers in general, we see AI everywhere, but the GDP growth statistics, right? Do you feel that your customers are approaching a tipping point where they are going to start to see faster growth that will be measurable? And if not, like, why not? Yeah, I think it very much is a customer by customer story. So I've got some customers, I think about a guy named Kevin Hearn, who runs software engineering for a

46:16bank called Access Bank. They're based in San Diego. They're an entirely digital bank. They have been at this for a long time. They are seeing huge gains in terms of their own productivity. And that has been the starting goal, right? Like dramatic acceleration of delivery of new capability for the bank so that they can better compete in a digital landscape for deposits, for customers, on lending,

46:46and so I think they're far ahead. I'll let them speak to their own financial results from that, but they've got a pretty exciting story. I think we've seen other examples of companies like this company, Yesco. I don't know if you've ever been to Las Vegas and the Welcome to the Las Vegas sign. They build these and operate these kinds of signs all over the world. And they have dramatically accelerated their productivity in ways that are exciting for them because it means that they can

47:18tackle more new business opportunity than they could with existing resources before. I'll have to let you talk to them about their own financial results from that. But they tell an optimistic story, which is exciting for us to be able to help fuel that. There is lots of potential. Everybody sees it. It takes a while in an organization for progress in productivity to translate into P&L results.

47:48And that's true in my business. That's true in any business. Everything doesn't happen overnight. And I think not all of the investments that people are making are going to pay off, right? There's a lot of experimentation here because a lot of these things are new. But I think everybody's pretty convinced that the potential is high. And we're going to see the returns as we learn how to do these AI workloads at scale. And I think that's the promise. And now, yeah, our job is to go help deliver it.

48:21So as OutSystems does go about delivering it, how much of the sort of savings of increased productivity do you think gets passed on to the customer? Like a lot of the sort of first thing is like we can be more productive. We have higher margin. Obviously, there's competition that in theory should discipline high margins. Where do you think we are on that story? And do you think that in the end post-transition margins are the same as they were before and everybody just gets a lot more for the money? Or do you see the structure of the industry changing in terms of who captures

48:57what value? I think it's going to be very different depending on industries. I think there are a lot of industries. I think of discrete manufacturing as an example where so much of the costs in the business are tied up in physical materials where the influence on what the margin profile of the business may be is probably a lot more closely related to tariff policy than it is to AI. That's just one extreme example. I take another example we all love to talk about, which is kind of the

49:29consulting industry, right, where I think we basically deal in information. And so in that world, I think there's a lot more where AI can take on more of the work and have a greater impact. And I think it will have an impact both in terms of like how much demand there will be for that kind of work and also the margin profile at which that can get delivered. So I do think it's going to be a very different story by industry and there's not one answer. I also think there's going to be a big

50:02variation within industries. Some organizations are moving fast, leaning in, trying things. Some are sitting back and saying, hmm, we'll watch, wait. And I do think that will have an impact on who gets there first, who will gain on margin, but who will gain on share as well. And therefore, who ultimately wins? When I tried to mentor a product that you described with the builder experience, I was struck by a couple of things. One was just, and I'm interested in your kind of take on what is the

50:36value of very early aggressive experimentation? Because one thing that has been true for me over and over again is I strive, I toil, I sweat to finally make something work. And then the next model generation or certainly two model generations later, it's that suddenly easy, right? So even building something like mentor two generations ago, probably pretty tough, right? To get it to work, to get it to be semi-reliable, et cetera, et cetera. Now you could put five, five or Astra on it and probably

51:09build something very similar to what you have in whatever, 10% of the time or even less. When you think about like strategically, what organizations should be doing, how do you separate spinning your wheels and just churning on stuff that will soon be easy from waiting? How do you know what things you really want to attack early and where maybe a little patience actually could be a virtue? Yeah. Okay. Good question. Lots to talk about there. Certainly like the build of mentor has been a big

51:40lift, right? Absolutely. We began this effort in 2018 before I was with the company. And obviously the state of technology, it was all machine learning at the time, there was no generative AI. It was back. The question was, how do you use AI or machine learning AI in every phase of the software development lifecycle to speed things up and make quality better and, uh, and everything that of course, the project has evolved a ton over that time to the point where now, you know, mentor is really

52:12a set of MCP services that can be consumed anywhere to build software reliably that, that always meets all of your non-functional requirements and helps you stick with your compliance goals and all those things. So like, it's been a huge transformation. The models we've used under the covers change all the time as we optimize for costs, as we talked about earlier, the problem that most of our customers still face today. And honestly, this is not a dramatic shift from previously, but capability is better.

52:43People can do more faster. But the real question is like, what should I build to actually solve problems in my business to take advantage of opportunities in my business? Like what to do is still the hardcore hardest question, right? So how much should I play around and experiment and try? I would say a lot of that effort should probably be going into where can I get the biggest bang for my buck? Where do I, where, what business process does it make sense for me to transform into something agentic? We've, um, we've learned the hard way on this with our existing customers. You know, we've got

53:18portfolios of thousands of applications today. So what we did is we built an AI that says, tell the customers what agentic system they should create over top of the assets that they already run on our platform. And we've just launched a brand new capability. It's currently in the hands of our

Experimentation and agentic workflows

53:36customer success teams, which is, uh, an agent that simply tells you what to build and forecasts for you the ROI in terms of time savings, all this stuff. We use all the telemetry data from the apps that you've built. We understand the structure of what you're doing. We see the data flow through those that we reason over that. And we say, Hey, you might want to think about an agent here for taking this manual job and turning it into an automated one. And by the way, press this button to build it and let the AI actually go build a V1 of that, that you can then, uh, run through, of course,

54:14your development, testing, validation process, and, and try it out. And so like solving the problem of what to do is like a huge part of the challenge for most organizations trying to get there in terms of an agentic transformation of their business. Very interesting. The other big question I had around mentor is what do you see as a future of the user experience in software? The mentor experience is the now increasingly familiar AI sidebar that kind of wraps the legacy

54:49application, right? What you used to do through drag and drop, it can now do through tool use. And now you have both, but then you also have this like MCP paradigm where I sit in some other software entirely, or maybe I even take a walk and just use voice mode to talk to my agent about it that way. And the, the tools are never perhaps even visualized in the way that they are on the core platform. What are you anticipating in terms of skating where the puck's going to be? Where are you trying to get ahead of those trends? Yeah, great question. I, I personally still

55:24imagine that a year from now I'm going to have more than a hundred apps on my phone, right? Like I do think that the user experiences for different things that we want to don't all lend themselves into a conversational experience. Like we think principally about whether we're doing it voice mode or text or whatever. Now I, I don't know how it evolves. I do think there's a real significant chance that a lot of software as we know it today is just widgetized into little things that show up

55:58in the flow of conversation, right? So like I think about the Google maps question. Google maps is one of the most useful things on my phone. I use it every single day, whether it's to look up a place to order a pizza or it's to navigate me to, you know, a destination where I'm driving or walking around a foreign city that I don't know well. So I'm like, okay, when does Google maps just become part of my conversation with an AI? Okay. Could happen immediately. Like I could just say chat GPT, give me turn by turn instructions on how to get from here to there. Um, but I still want to know

56:32like, what the hell does the world look like around me? How is it shaped? What am I going to pass on the way? And I don't think the conversational experience is going to lend itself perfectly there.

56:42Microcosm of, you know, the whole challenge you're putting forward here, but an example of where I think things will still be different. I think for a variety of reasons that could be oriented around privacy, that could be oriented around compliance, that could be oriented just purely around user preference and experience that we're going to still see a wide variety of software interaction paradigm in the future. Now I'm open to the idea that the big monolithic chunks of that, that we have

57:17today, a portal I go to with 80 features might be delivered very differently in little chunks, just when I need them in the flow of other work. And, you know, that's going to be interesting to watch and see you develop. I think that we, we at our user conference in June, we showed a pretty compelling demonstration of one of our banking customers initiating loan applications through a consumer conversation with Claude about a home remodel. Hey, let's think about this new stove and all

57:53these things. And like, Jesus is going to be expensive. I might need some money. Do you know you're pre-approved from a loan for a loan for 30 K? Would you like to initiate that application? And a seamless flow from the conversation in Claude straight through to the mobile app of the bank with the same agent operating behind the scenes in both contexts, seamlessly carrying the conversation through, recognizing which typically required assets you need for a loan application process that they already have,

58:25because you're already a customer of the bank. They don't need to ask you for your bank statement and straight through processing that kind of application through the backend where there's agents doing the work or making recommendations to humans in the loop at every phase of that process and streamlining the entire experience for a consumer. That's interesting, right? There's, you know, a lot of the workload moves directly into the, the agent experience, um, if you will, but there's

58:56still some software assets there in this case, in the form of a mobile application that need to be there for a series of compliance and, and privacy and regulatory requirements. Will those change and evolve over time? So that's not needed. I don't know. We'll see. I think the interesting question for me is kind of not how are all the things going to become the same, but what will the imagination now of creators produce now that AI and platforms like OutSystems remove so many of the typical traditional

59:28constraints of building amazing experiences. That notion of everything becoming the same and also your comments earlier around how many major features you guys have been shipping recently and how much that's grown has me wondering about the future of competition. I do feel like everybody is newly emboldened to go chase these adjacencies and the more adjacencies they chase, then the more new adjacencies become available to chase. And so I do see this convergence of like all the platforms growing

1:00:03into the same super horizontal, everything you need in one place kind of offering. And I wonder if you see that too, and if that is where you think it's going, how do we win? How does one win in that environment?

Specialization in a crowded software market

1:00:16Yeah. Yeah. I mean, of course, let's be clear that if you go drive the one-on-one freeway in San Francisco and you see all the billboards and they all say the exact same five words on them, right? Like it's just a different logo. And for sure, I think the industry is struggling to crystallize differentiated messaging in this world. And you know why? Look, part of it is the sea of same, right? Like so many experiences across so many dimensions today can be delivered through a

1:00:47conversational experience. And that's kind of one factor. It's what's behind that conversation that now is the question. How, what are the systems back there? What is the data back there? How do those things interact back there to give me more of what I want through such an experience? And, you know, you're like pretty quickly, you come to, well, okay, great. I need an agent platform. I need an AI builder. I need integrations to lots of things, right? Like the stuff it takes to deliver on that

1:01:17question is kind of some basic building blocks, right? And at the moment, a lot of platforms are rushing into that space and saying, okay, well, let's fill it up with all those things. So how will we differentiate in the future? I think it's going to be really, ultimately, it's going to be specialization, right? There's going to be a common set of building blocks everybody needs. And then there's, what do you get really good at? What do you optimize for? And part of that optimization will be in domains like tuning and distillation and model specificity and associated cost structures that

1:01:53optimize for that. I think part of that will be, you know, in industry and workflow specific optimizations, I think part, and you know, there's going to be concentration around that gets you out of just pure horizontal. Everybody has the same stuff because everybody has the same primitives because you need primitives, but you got to have specialization on top. So for us, you know, a lot of that is in regulated industries where we've solved a ton of those problems in ways that we can bring to bear and to this platform. It's like the 37 features that are not on that basic list that you need

1:02:26to actually get through the hurdle of building or deploying the first thing. And so I think that's going to be a big part of it in our domain. And then we've got specialization that we've built over years in banking and insurance and in government and healthcare, transportation logistics, energy and utilities, which are all big regulated industries, all of which are the core of our installed base today, where our teams have a lot of specialized skill and understanding and where we can rapidly

1:02:57help organizations modernize processes. And so I think it's going to be that kind of specialization, how we differentiate. At the moment, we're in the thing, we're in the point where everybody's saying, yeah, we got all the stuff. And I think we're all rapidly learning that that's just confusing everybody. Everybody's like, oh, everything's the same. And so I do think we have work to do as an

Hiring junior talent and onboarding agents

1:03:21industry there to get our game on. How have your hiring practices changed? And what advice would you give to people who are just entering the workforce these days, who I think got a lot of signals recently that they're nervous about their own employment prospects? They're somewhat resentful toward AI because they feel like it's competing with them. What are you doing at AdSystems and how would you translate that to broader advice? Yeah, I'm super bullish on junior talent, on people who grew up

1:03:51whose last five years have basically been native to all these new technologies where they can bring those in to organizations, our customers, our partners, our own organization, and really drive accelerated innovation. They obviously need to learn a lot about the core business processes and all these industries to be of value. But I think that they bring generally a fresh perspective and attitude and AI first, AI-pilled, if you will, kind of approach to things. And so I'm bullish there.

1:04:24I don't think the organizations today have built the infrastructure to do the other part yet well, which is really indoctrinate these people into the industries and the specialized knowledge that it takes to put those skills to use as effectively as possible. What's changing about the way we're hiring? We're definitely leaning into the FDE model, thinking about how do we help customers more directly in imagining the work, right? A big part of what it is, like designing, where are we going to

1:04:57make the investments to have significant impact? And we think about that as the people we hire, I was thinking about the agents we hire. And so the work I was talking about today about our agent foundry that goes in and inspects your architecture and says, here are opportunities to turn into agentic is another thing that we're doing. So we're thinking very differently today about every new piece of work that needs to happen. And is that work for a human? Is that work for an agent or a team that's a blend of those? And how do we structure everything around this new model for delivering

1:05:32work? And then we're thinking about the journey of every employee that we hire and how we dramatically accelerate what we've typically called onboarding, which, you know, for decades in most businesses was like, hey, here's the new hire thing you do for a couple days or a week. And here's the follow-up trainings you do on all these things. And shifting all of that into something that is more agentic, more real-time, more just-in-time, give you the knowledge you need at the point when you can actually then use it and therefore internalize it. And so a lot of the practices around people

1:06:07are evolving fast. And those are investments that we're making in our own right. And we're certainly encouraging customers to think through in their businesses and how it fits there too. Anything else you would want to leave people with before I let you get back to work? Yeah, well, tune in today to the OutSystems World Tour from Las Vegas, where we will be showcasing a lot of these amazing technologies. The latest in the agent experience for OutSystems, which dramatically

1:06:37helps organizations to accelerate trusted delivery of AI systems for their business. Woodson Martin, thank you for being part of the Cognitive Revolution. Cheers, Nathan. Cheers, Nathan.

1:06:50Someday came. Someday came. Oh, someday came.

1:07:01Faucet dripping all year long. Guitar waiting on a song. Seeds in a jar, a letter unsent. That's where all my Sundays went. Woke up laughing at myself. Half the list didn't sell. Someday came. Someday came. Rang the bell and called my name. Brought more friends than I've ever known. Someday came.

1:07:33And it ain't going home.

1:07:38Throw the windows open wide. Let the whole bright world inside. Let the whole bright world inside. Every song I meant to play. Every friend is on the way. Pinch me, tell me I'm awake. Every dream's a piece of cake. Someday came. Someday came. Put every maybe to shame. Moved right in with the key of its own. Someday came.

1:08:08And it ain't going home.

1:08:14Used to wish on a falling star. Now I say it and there you are. Used to save it all for when. Now it's why not two? Why not ten? Home in the daylight. Time to play. Nothing on the list says not today. Someday came. Someday came. Sing it out and shout its name. Every seed I saved is grown. Someday came.

1:08:45And it ain't going home. All those years I said someday, someday was no in a kind of way. No moved out and the sky moved in. Say the word when we began. Someday came. Someday came. Rang the bell and called my name. Brought more friends than I've ever known. Not alone. Someday came and it ain't going home. Tell them all.

1:09:16Someday came. Someday came. Someday came and it ain't going home. If you're finding value in the show, we'd appreciate it if you'd take a moment to share it with friends, post online, write a review on Apple Podcasts or Spotify, or just leave us a comment on YouTube. Of course, we always welcome your feedback, guests and topic suggestions, and sponsorship inquiries, either via our website, CognitiveRevolution.ai, or by DMing me on your favorite social network. The Cognitive Revolution is part of the Turpentine Network, a network of podcasts, which is now part of A16Z, where experts talk technology, business, economics, geopolitics, culture, and more.

1:09:54We're produced by AI Podcasting. If you're looking for podcast production help for everything from the moment you stop recording to the moment your audience starts listening, check them out and see my endorsement at AIPodcast.ing. And thank you to everyone who listens for being part of the Cognitive Revolution.