Three years ago I left a fifteen year career to bet on AI. This is what I found.

I am disappointed. Severely disappointed. As a matter of fact, my whole 35 years of existence has been one long string of disappointments, and I think it is time I talked about it.

We are in the fourth year of the supposed AI revolution. And yet... where are all the machines? Where is all the automation? Where is the transformation that was supposed to be so profound it would reshape the very foundations of how we live and work?

But I'm getting ahead of myself. Let me back up.

As a child of the 90s, I grew up on a particular brand of optimism that I think only people from that era truly understand. The Cold War was over. The internet was coming. The adults in the room were telling us, in so many words, that the future was going to be better. That the arc of history bent toward progress. That we, the generation inheriting this brave new millennia, would have it better than anyone who came before us.

Instead, I got 9/11 when I was ten. A financial crisis when I was seventeen. A pandemic when I was twenty-nine. And somewhere in between, the slow, creeping realization that what I was actually living through looked a lot less like progress and a lot more like civilizational decline. Either I was cursed and doomed to suffer through interesting times, or I live in a simulation being run by someone with a very dark sense of humor. In any case, it was not what was promised.

But of all the disappointments, and there have been many, the one that stings the most is AI. And I say this as someone who has spent the last fifteen years obsessing about it. I was never supposed to be in tech. The original plan was finance, and there is likely a universe out there where I went down that path and right about now would be fairly stressed about AI capex numbers. But as fate would have it, I read a book in business school called "The Singularity Is Near." It was a long and tedious process to get through, and most of the myriad of technological breakthroughs discussed in that book are lost on me now, but AI stuck. And by extension, the implication it would have for economics, which to me was far less interesting than the technology that had the potential to make the concept of economics as we understand it obsolete. That thought never left me. So here I am, fifteen years later, three of which have been spent building around it, thinking about it, and watching it more carefully than most.

I remember when breakthroughs in AI were few and far between. I remember the GO moment, when DeepMind's AlphaGo beat Lee Sedol, and for a brief window it felt like the future had cracked open just a little. Those moments were rare, and they were special because they were rare. Now I am swimming in hype. Every week there is a new model, a new benchmark, a new breathless announcement from a CEO who has confused fundraising with scientific progress. And herein lies a problem that I, dear reader, would like to discuss with you today.

Some of you will say this is a controversial claim. Look at ChatGPT, look at Claude, look at all the tools that didn't exist three years ago. And you would be right that something has changed. But here is the thing about all of these tools, and I need you to really sit with this for a moment: at their most fundamental core, these are collections of text files containing instructions and commands, hooked up to a messaging app. I don't know about you, but that doesn't exactly sound like the singularity I read about in business school. Yes, I'm being reductive. But sometimes reductive is clarifying. Strip away the interfaces and the branding and the billion-dollar valuations, and what you have is a system that predicts the next word. That's not a criticism. That's a description. And the gap between that description and what's being promised is what this series is about.

I'm not saying it's nothing. I'm saying it's not what was sold.

And this is where my story gets a little personal, because when ChatGPT first landed in late 2022, I was working at Microsoft. Given the partnership with OpenAI, I had a front row seat. And I could feel, in a way that's hard to describe to anyone who wasn't in the room, that something fundamental had shifted. By mid-2023, I made the decision to leave. Not just Microsoft, but just over fifteen years in the corporate world entirely, to go do my own thing.

I had no idea where any of this was headed. Nobody did, but at least I was honest with myself about it. What I did know was that I needed space. Space to research, to study, to build the correct lens through which to look at this technology before I started making bets with it. And more importantly, I needed to build a team.

What I didn't fully appreciate at the time was what "space" would actually cost me. I went from a well paid job in the corporate world to doing grocery deliveries in Amsterdam. By bike. And by truck. If you live in the Netherlands, you know exactly which small electric trucks I'm talking about. In Amsterdam, truck delivery is more like a six hour crossfit workout, and I should know, I've done both. Stop and go. Stop and go. Stairs everywhere. And just sheer work in between.

I'm not going to go as far as saying it was fun, but I met many new and interesting people during that time. There is no bond quite like the ones forged when you are suffering together. And it gave me an insight into what life at less than twenty bucks an hour actually looks like these days. I opened this entry talking about how disappointed I've been, how hard things have been for my generation, the broken promises of the 90s. But working those jobs, alongside people younger than me, I realized I can't hold a candle against the generations that came after. Whatever disappointment I carry, theirs is heavier, and they didn't even get the luxury of the promise first.

That was the hardest stretch of this journey by a wide margin, though not the hardest of my life. I did all of it to make ends meet while I went back to "school" on AI, dusting off everything I had accumulated over the years and reorienting it toward something that was genuinely new. Because working in "Data and AI" in tech before 2022 was a lot more Data than AI, by a wide margin. The most advanced stuff most of us were doing was simple regression or decision trees using XGBoost. What had just arrived was something more primal, something closer to the core of what AI was always supposed to be, and I know for a fact it caught most AI researchers at the time off guard, let alone those of us on the applied side. I needed to build the right lens before making a single move.

And through all of that, I was living in this duality that I think a lot of founders in this space will recognize. My gut was telling me to wait. To observe, assess, plan, and then execute, a methodology I learned early in my career that always maximized the chances of success. But my brain was screaming at me to jump in, because the opportunity was immense, and I had so many ideas for how this technology could be deployed. The temptation was enormous. But I also knew, deep down, that I didn't understand what I was looking at. Not really. And I am not personally one to ask for money when I don't know what I'm asking for.

So the gut won. I chose to observe first. It cost me. Everything got harder, and fast. While others were raising money on decks full of promises and hockey stick projections, I was sitting on the sidelines, watching, reading papers, testing models, trying to understand the shape of this thing. It felt like the wrong decision many times. It felt slow. It felt like I was falling behind.

But three years later, I have something that most of those founders don't: I have a report. And it's not what either side of this debate expects. It's not the doom story, and it's not the utopia story. It's something more interesting, more nuanced, and honestly more useful than either of those narratives.

But I have more than a report. That report creates a situational awareness, a foundation to build on. And because I chose to wait, because I chose not to take money before I knew what I was asking for, I have something else that the vast majority of founders in the AI space do not: I am unrestrained by past promises of returns. No investor deck to live up to. No hockey stick projection haunting my quarterly board meetings. Just a clear picture of what this technology actually is, and the freedom to build accordingly.

Over the next few entries in this journal, I'm going to share what I've found. Where the money went. Why the technology has a ceiling that most people aren't talking about honestly. What this thing actually is, as opposed to what it was sold as. And what I think the real opportunity looks like for those willing to see it clearly.

Because at the end of the day, this isn't just about technology. It's about whether this thing can actually deliver on the kind of promise that keeps getting broken, for my generation, and for the ones that came after who never even got the promise to begin with.

A Journal from the AI Frontier: The Disappointment

Stay in the loop

Ready to Get Started?