Charles Fry, Founder and CEO of CODE Éxitos, is helping businesses hire AI in enterprise to transform software engineering, modernize product development, and build intelligent connected systems.
In this conversation, Charles introduces The Agentic Org Chart Framework: Hire Systems Thinker, Look for Failed Entrepreneurs, and Have AI Replace “Trade Skills”. He explains why AI is fundamentally changing software development, how organizations must redesign their structures for an AI-first workforce, and why systems thinking is becoming more valuable than technical specialization. Charles also discusses how the rise of agentic software development is reshaping the future of SaaS, why combining AI with connected hardware creates a stronger competitive advantage, and what business leaders must do to successfully navigate AI-driven transformation.
—
Hire AI in Enterprises with Charles Fry
Good day. Steve Preda here, and I’m talking with Charles Fry, the Founder and CEO of CODE Éxitos, building cyber-physical systems for mid-market and enterprise companies, as well as full-stack web, mobile, and SaaS development. Charles, welcome back to the show.
Hey, I’m happy to be here. It’s always great to see you, and I’m looking forward to our chat.
Yeah. It’s so interesting to talk to you because the last time we had you on the show, three or four years ago, it was still before the AI age was fully upon us.
Right.
Your business was kind of a different business. I’ve been following you on LinkedIn, and I see that you’ve evolved your approach, and now you’re an AI-first company. So tell me a little bit about how that came about, this whole evolution, and how you found your new focus?
Yeah. Wow. It’s been that long since we were on the show. It was a lot of fun, but here we are. You’re right. AI really sort of came out of left field. I’ll skip all the technical things that suddenly made AI an achievable thing. But really, at the beginning of 2024—and somebody can fact-check my timeline—ChatGPT, if you were aware of it, was kind of passing the Turing test. It was giving pretty reasonable answers to natural-language questions.
And we were like, “Wow, this is interesting.” At first, we were helping our clients think about how to build those capabilities into their products, something we still do. Share on X But by mid-’24, late ’24, it became pretty obvious that one of the best applications of large language models—and expert systems, we used to call them that—is writing software. And so by early to mid-’25, the systems were suddenly not novelties. They were credible at what they were doing, and they were gaining momentum in the quality and credibility of the software they were producing.
Now, CODE Éxitos was started largely to create an opportunity for entrepreneurs and enterprises that needed, let’s call it, garden-variety, well-done software. We built that through the Americas to arbitrage labor rates in Latin America and leverage the time zones. So it was essentially an offshore, blended, hybrid-team model. Honestly, by the middle of 2025, the AI tools for software engineering were as good as, or better than, 50 percent of the human developers that we employed.
And at some point, as a business owner, when you’re out there representing yourself, and your product is yourself, and you’re representing that to clients, you have a moral and ethical obligation to say, “Hey, I think I can still give you the best product that you’re looking for, but I’m going to do it in a different way.” So beginning in late 2025, we were hard into the pivot. Today, all of our software development is done agentically. There are still people. It’s not a complete dark factory. But our mid-level performers and below, we exited them from the business, which caused a lot of human turmoil. I mean, we were at around 100 people. A lot of human turmoil. Our clients were going through the same thing. We were watching what was happening in their organizations and what the leadership demands were.
We just made the decision to lean into it. And here we are, almost mid-’26 now, and it’s actually going really well. Now, AI, of course, anyone who opens an internet browser anymore sees it. AI is everywhere. You read the newspaper. AI is going to change everything.
Every application has an AI layer to it.
Yeah.
Right? Every SaaS application has a button that says, “Use AI here.”
My view, at least—and these are the things people should probably give me some credibility on—I’m going to keep my views focused on how AI applies to the software industry, my industry, and its direct impact. We see, and we have clients working on, things like AI in customer service, the legal department, and the finance department. We do all of those things internally—agentic first, AI first. But those really aren’t my industry. I’m not ready to make a sweeping prognosis about how AI is going to change capitalism in the United States. But in the software industry, it’s a fundamental change. And it’s not done yet.
So what’s your vision? Where is everything going? What’s it going to look like three years from now?
Well, I’m not smart enough to know that. But I think the patterns we've seen—and again, within the world of writing software, and making that a broad category of activities Share on X —are going to continue to consolidate and converge to where humans are important, but they might be only 10 to 20 percent of the input to the process. I really do think we’re going to see a day, sometime in the three-to-five-year time horizon, where a large amount of software will be written and managed by other software systems.
There’s no technical reason to prevent that. For example, I was talking to one of our clients today, a CIO at a great company. A couple hundred million dollars in revenue. A really well-run business. A sizable internal IT team. But there’s a lot of ongoing maintenance and attention required. Building the software is just the beginning of a five-to-ten-year life cycle. So I think in the near term we’re going to see that building the software becomes, “Okay, we got that figured out.” That’s a largely solved problem.
We’ll then progress to the problem of: “Hey, this software has been in production for five years.” “It needs updates.” “It needs attention.” “It needs maintenance.” “It needs to scale.” More software systems will take care of that. Fewer and fewer humans will be involved in that kind of work. So I think that’s where the software industry is headed. I think it’s going to be 60 to 80 percent smaller in human capital than it is today. Sometime soon. Yeah. I really do think it’s about as close as I want to get to calling it an extinction event. Let’s put it that way.
Some people say SaaS companies are going to go out of business, and it’s all going to be agents running around doing things for us. But other people say SaaS companies are actually the SOP for whatever activity is out there, and you need that structure. The SaaS application provides that structure. You don’t want agents running in an unstructured way. You’d rather have these SaaS applications. What’s your view?
I think that’s a good way of looking at it. If you’re a dinosaur like I am, back in the late 1980s or early 1990s, when you wrote software for a company, everything was custom software because there were no packaged software products, no SaaS platforms. But over the last 20 years, I think SaaS companies have become, for a lot of businesses, exactly what you said. They’re the embodiment of best practices. If you take something like HubSpot, which I’m sure you and many of your audience are familiar with, you really don’t need to customize it.
You just need to follow its baseline processes because they have thousands of customers who have helped refine the sales motions that work. So I think there’s some truth to that. The problem SaaS systems face is that the barrier to competitive entry is much, much lower. If you look at a company like Salesforce—and I think I’ve led three different Salesforce deployments back when I was a CIO or CTO—that software really shows its age. It’s layers and layers of complexity built to serve a wide audience. It’s great. It’s expensive. Emerging companies don’t need that. They can effectively vibe-code their own CRM system, and it works. I think the threat for big SaaS companies is twofold.
One is that the next generation of customers is going to onboard very differently into those systems than the previous generation. Share on X I don’t know what that onboarding ramp is going to look like. The second problem is there’s very little defensibility in having a pure software product. And the other part of our intro—you talked about cyber-physical systems. We’re spending more and more of our product development cycles on hardware-related products, things that have a nexus in the physical world. Here’s a good example. I’m wearing one of these health rings. This Oura Ring.
Oura, yeah.
There’s a lot of amazing hardware in here that justifies my monthly subscription for the app. The app we could recreate pretty easily. But the development, manufacturing, and distribution of this physical item create a much higher barrier for a competitor to overcome. So more and more of our clients are companies that have a physical product they either want to make smarter or make more connected.
That’s really what it comes down to. And that’s a pretty exciting space. But for a pure-play SaaS company, I think it’s going to get tough. The competitive pressure is going to be intense. And the barrier to entry is going to be really low. It’s not even about engineering cost anymore because the cost of engineering has dropped so much with AI.
It’s almost like it went full circle. You had all these product businesses that wanted to become service businesses to create recurring revenue. And now the service businesses—the SaaS businesses—want to become product businesses to create a barrier to entry, improve retention, or reduce disruption. Isn’t that interesting?
Yeah. I hadn’t thought about it exactly that way. Maybe the pushback would be that these professional services businesses wanted to have a technology play or a platform. That’s interesting. But I think we’re going to see AI, at least in technology, continue to lower the barrier to entry. It will allow much faster experimentation with pure software ideas. And we’re focused on the things where the AI robots can’t play. They’re not going to cut your grass.
They might guide the machine that cuts your grass, but they’re not going to cut your grass. So I think that while the turmoil is still sorting itself out in the pure software world, we’re going to see a whole new set of opportunities open up. We’ll be able to build truly smart devices. Devices that think for themselves. Devices that are aware of the world around them. They can participate with us in our day-to-day work. That’ll be a lot of fun. I think we still have some rough sailing ahead of us as AI sorts itself out.
Isn’t it true that people prefer to interact with a purpose-designed device rather than a software product? And maybe an app is kind of a device that is software, or maybe that’s the overlap there. But I know there are some things I’d rather have on my phone, even though it’s complicated because there are so many other things on it. But if I have a single-purpose device, like you have your Oura Ring, it’s easier to interact with. There’s no complexity, and then it lowers the accessibility.
Yeah. An area of active study is something called HMI, or Human-Machine Interface. Again, back in the ’80s and ’90s, it meant things like: Are the buttons big enough for someone to push? Does a red light always mean a bad thing, and a green light always mean a good thing? But now that’s expanded into the kind of research in psychology and sociology that you’re talking about, Steve. Some of that is really amazing.
I’m sure you’ve seen them, and your audience has seen them. You can find these videos on YouTube. There are humanoid robots. It took a while for researchers to figure out that a robot doesn’t actually need a head. It can have what is essentially a torso with arms and legs. The head doesn’t really need to be there. But a robot with no head freaks people out. Yes. People don’t like it. So the robotics engineers put heads on them. Then they thought, “Well, if we’ve got a head here, we’ll just put a face on it.” But if the face is too realistic, it gives people the creeps.
Yeah.
So people didn’t like faces on them. If you look at the current generation of humanoid autonomous robots, they have these, I don’t know, sort of pseudo-faces. They kind of look like Halloween jack-o’-lanterns or something. They’re not scary, but they’re somewhere in the middle. Anyway, the things you’re talking about are really fascinating.
I mean, Isaac Asimov wrote about humanoid robots and all the challenges that come with them. What happens when they have a human-like appearance? What happens when people think they are actually people, but they just don’t age? All those things have been explored. But listen, I’d like to switch gears here and ask you this. Right now, in this AI age, what drives your business? What drives growth in your business?
This part is truly fascinating to me. Everyone is working off the same timeline now, which isn’t a very long timeline. We don’t have a lot of experience to draw on. Much more quickly than when the internet became commercially available—I was there when that happened too— the adoption of AI as a fundamental change happened in a matter of months, compared to several years for the internet. Share on X Some people also compare it to the adoption of mobile phones, which you mentioned.
But this has happened very fast. A year ago, we were talking to sophisticated technical buyers who said, “Yeah, I’m still on the fence about whether I like agentic software development.” That doesn’t happen anymore. Everybody says, “Yeah, we’re using it too.” What we’re seeing now is that it’s evolved so quickly and had such a fundamental impact that people don’t know not only how to manage it inside their business, but also how to deploy it. It’s really disruptive. And this is where you’re a pro.
It’s really disruptive to organizations. So in less than a year, we’ve gone from, “Hey, should I let my developers use AI?” to now everybody using AI. And the leading teams, including ours, can produce high-quality commercial code faster than organizations can absorb it, and faster than org charts can adapt to the change. Our engineering teams have to adapt to the pace of the business, not the other way around. Because we get clients saying, “Hey, you guys are producing too much.” “We can’t check everything.”
“We haven’t finished testing last week’s new features and capabilities yet.” “We can’t take another batch of features and capabilities this week.” So we’re seeing a lot of organizational behavior change starting to come out of this. When we’re talking to C-level executives and senior leaders, that’s where most of the conversations are today. “How do I retool my organization to capture the benefits?”
Yeah. Because what I see is that the more AI you apply, the faster decision velocity becomes. And the complexity of understanding the whole picture, connecting the dots, increases. So you need a different kind of person who can operate at that higher level of contextualization. Do you see the same thing?
We do. Software engineering and product development had matured into a pretty predictable set of job descriptions and capabilities. The business processes were really well worn. We knew what the product owner did. We knew what a project manager did. We knew what a tech lead did. Et cetera, et cetera. A lot of these, frankly, became trade skills. “Hey, I’m really good at writing code.” Or, “I’m really good at doing QA, but I’m not really a systems thinker.” “I’m not an entrepreneur.” “I’m not a creator.” Pick your fuzzy lens of choice.
That’s really what AI displaces right now. AI displaces those trade skills and, frankly, does them better and cheaper. There’s no way to dispute that. What we look for now, and I think where the trend is going with our clients, is systems thinkers. We have enough agentic tooling built on our own internal platform that the people operating and building products for our clients Share on X —we refer to our team as digital creators—come from a variety of backgrounds. You don’t have to be a computer science major. You do have to have some domain experience. You do have to be a systems thinker. You do have to understand what business value you’re trying to create. But as far as actually writing really good code, nobody’s really doing that now. It’s being done automatically.
If you have a couple of gray hairs like I do, you’ll remember back 20 years ago when we talked about the war for talent. That’s what everybody was looking for. They wanted people with these highly specialized engineering skills. I think there’s a new war for talent. It’s going to be harder to pin down because we’re going to be looking for whole-systems thinkers as opposed to technical specialists. Because AI will be the technical specialist we need, regardless of the business domain.
So what do you do to infuse systems thinking in your business?
Wow. I wish I had a good answer for that. I don’t even know how to recruit these kinds of people right now. I’ll be that candid with you and your listeners. I was talking to a couple of my senior people, and I said, “Maybe we should go look for failed entrepreneurs.” Which is kind of a heretical thing to say. But as an entrepreneur, I know firsthand that it doesn’t always work. The fact that the business fails doesn’t necessarily mean you, as an entrepreneur, are a personal failure.
Entrepreneurs are about the only, I don’t know, primary source I can think of for people who have done a little bit of everything. They’re systems thinkers. Maybe they didn’t get it right, but they could. So we talked about that. I don’t know if Disney still does it, but Disney was phenomenal at producing these kinds of people through its internal training programs. We’re not big enough to compete with Disney. But to answer your question, how do we teach it? I can’t honestly say that we do. Because we’re still figuring out what it is that we would even teach.
Well, it’s a new type of SOP that’s needed in the business. So what are the best practices for building an AI workforce? That still needs to be defined.
Yeah. For larger organizations, our clients that are running larger organizations have the same problem. All of a sudden, their org chart is broken. What I mean by that is, if you look at the way we’ve traditionally built and scaled businesses, you have this pretty large cadre of managers who give you what Eliyahu Goldratt called the span of control, your degree of leverage. When you’re younger, you hear things like, “Ah, my manager doesn’t even do anything.” You’ve probably heard that before. “Oh, my manager doesn’t really do any work.” “He just comes in and bugs me.”
It’s not entirely wrong because we rely on that manager’s experience to be spread across six, eight, or ten other people and supervise their work. So managers don’t really do a whole lot of delivering the work themselves. I think AI is going to change that. I know AI is already changing that inside technical teams. All of a sudden, we have clients saying, “My org chart doesn’t translate to the way my business operates under this new agentic model.” That’s problem number one. Problem number two is, “I have people on my team who are good people and good contributors, but there’s no box for them in the new org chart that I think works with an agentic workflow.” Does that make sense?
Yeah.
So two things have happened suddenly. We see a lot of press—although I think it’s moderating a little bit now—about how kids coming out of university are having trouble getting entry-level jobs. True enough. I think the other area where org charts are collapsing is managers who don’t actually produce any output. Supervising other managers and compiling the weekly report of reports just isn’t a valuable job function, even at the best of times. And now—I hate to say this—but it’s useless. So I think there’s a lot more work we’re going to have to do with our clients, and ourselves, on what an org chart should look like. And what the expectations are for people doing work inside a company that’s moving aggressively toward leveraging AI capabilities in what we would normally call white-collar job functions.
Yeah. Some years ago, I thought about the org chart being broken. The top-down hierarchical org chart—I think it’s completely broken. In my head, the org chart is more like an amoeba, where you have the entrepreneur and the manager in the middle as yin and yang. Different departments report to different people. And you’ve got these pizza teams all over, actually delivering teamwork. But that’s fascinating. Yeah, fascinating topic. So who are the ideal customers for you? If they have the right kind of projects, what are the right kinds of customers and the right kinds of projects for CODE Éxitos that would be interesting to look at?
There are two types of clients that we focus on and that we can help quite a bit. The first type of client is one that has a large, established internal software development and product development process, and they’re trying to figure out how to adapt, modernize it, and harness AI. We can come in, and we have a very opinionated point of view. We can have a couple of conversations, and they either like the direction our telescope is pointed in and want us to help, or they say, “No, we think we’re going to do it a different way.”
And that’s okay, too. Because right now, nobody really knows the final answer. We call those engineering transformation projects. Somebody says, “Hey, we have a team. Can you help us get better?” The second type of client we like is one that has this physical connection challenge. I’ll give you an example. We have a client that primarily makes pumps and motors. They put those pumps and motors into very specific industrial applications across North America. They wanted those pumps and motors connected to a network.
They wanted to collect data from those pumps and motors to help their customers. Once we built the data collection, the electronics, and the connectivity, the data started coming in. Now there are a lot of AI-related things we can do with that data. We’re beginning to work on what you can think of as supervisory agents that watch what’s going on. They’re much more robust than the old filters that just looked for exceptions and red lights. Those are the kinds of clients we really help on the product side.
Sometimes they come to us with an idea scribbled on a napkin. It’s like, “Hey, we have this system or this process that we envision, and we need somebody to help us build it.” We’ll do the electronics, the AI, and the software engineering. And that becomes a complete system. So, more systems thinking.
Yeah. And then you combine software with hardware. Then you have AI agents doing much of the coding, management, and maintenance of these systems.
Yeah, that’s right. This is not a client of ours, by the way, but the story I’m going to tell is fascinating. It’s a second-degree connection that I chatted with. He runs a $100 million-a-year manufacturing company. I think he’s second or third generation—I can’t remember which. He knows the business. He grew up in the industry. He’s in the process of transforming the company so that he can basically run the whole business from his phone.
He’s applying AI to his internal business functions like finance and accounting. He’s done some really amazing stuff. He’s automating the manufacturing process, the machines, and the feedback systems. It’s just stunning what this guy is already able to do. He just picks up his phone and says, “Yeah, I want to run my company from here.” I think he’s going to be able to do it. I think we’re going to see more of that emerge.
Yeah. That’s fascinating. The race is for the first one-person unicorn, right?
I think that’s already happened. I don’t know if you’ve seen this. I can send it to you. There was a New York Times article a couple of months ago about a guy—not a tech guy, a marketing guy. He and his brother run an online company that generates $1.8 billion a year in sales. And it’s just the two of them.
Wow!
Spoiler alert: As I remember, they sell weight-loss drugs. He’s a marketing guy with a tech background. He figured out all these marketing channels where, if you order Ozempic or whatever online, it basically drop-ships from the pharmaceutical company to you, and he gets a cut. But he said in the article, “At one point, we were doing $300 million a month in sales.” He goes, “Well, technically, I’m not a one-man company because I had to hire my brother to help me out.” So it’s the two of them.
That’s pretty great. Yeah. That’s definitely a unicorn. So if you’re listening to this and you have an enterprise company, and you want to harness AI in your business, create agentic systems, streamline your operations, and make your company more efficient and productive, then reach out to Charles Fry at CODE Éxitos. Any last thoughts for listeners who are thinking about building a business in the AI age? What advice would you give them?
I don’t know that it’s changed a whole lot. Building a business is always hard work. For any listener who wants to chat about any of these topics or see if we’re the right fit, they can always reach out and contact me. The conversation is free, and I usually learn something from it. But no, I would say that things are different. It doesn’t mean they’re wrong or better. I think they’re just different. It’ll be interesting to see how all of this unfolds over the next few years.
Yeah.
I’m in it for the journey.
We’re living in interesting times. So, Charles Fry, Founder and CEO of CODE Éxitos, thanks for coming on the show. And if you enjoyed this show, make sure you follow us on YouTube and Apple Podcasts. Give us a review, and stay tuned because every week I bring an amazing entrepreneur onto the show. Thanks for coming. Thanks for listening.
Thanks, Steve.
