I Didn't Plan Any of This

If you'd told me in February this year that by September I'd have shipped twenty-six pieces of software, trained a language model from scratch, and taught a security curriculum to my own engineers, I'd have laughed and gone back to whatever fire I was putting out that week.

I run delivery for a bank. My last hands-on coding was mainframe work, a long time ago. I don't say that to be modest — I mean it changed how I think this whole thing started.

It actually started on a phone. In a hospital room, of all places. Fifth of March, if you want the exact date — I do, because I went back and checked the file timestamps later, mostly out of curiosity about how far this had actually gone.

No laptop nearby, nothing planned. I'd been reading about AI agents for months the way most people do — articles, LinkedIn posts, the occasional demo video — and I finally got tired of reading about it. I pulled down a self-hosted agent stack just to see if it did what people said it did.

It did. Badly, at first. But it did.

What I didn't expect was how that one evening would refuse to stay contained. A week later I'd built a six-node agent workflow for interview prep, mostly to solve my own problem. A few days after that, an Excel chatbot, built in one sitting because I got annoyed re-explaining the same spreadsheet to three people.

None of this had a name yet. It was just a habit that hadn't figured out it was a habit.

The first thing that felt like a real product happened by accident. I'd been prototyping an org chart tool in a chat conversation — the kind of back-and-forth you do when you're bored on a call — and somewhere around day twenty I looked at what I had and realized it actually worked. Drag-and-drop hierarchy editing, PDF export, the works. Six days from idea to something I'd actually use.

That was the moment the hobby stopped being a hobby.

I won't pretend the next five months were smooth. There's a version of this story where I make it sound like a highlight reel — twenty-six products, sounds impressive when you say it fast. The honest version has a lot more 11pm debugging sessions than that sentence lets on.

I built a testing platform to replace a Postman/ReadyAPI setup that was costing real money, and its "why does this return zero results" bug took me three evenings to actually root-cause — turned out I was reading a bare results array instead of the full response envelope. I trained a tokenizer from scratch before I ever touched a proper fine-tune, because I refused to just call an API and call it understanding. I shipped a native Android app with genuinely zero prior mobile development experience, which in hindsight was a slightly unhinged decision on a Tuesday.

What actually made it possible wasn't talent, and it definitely wasn't free time — I still had a full delivery job carrying four markets the entire time. It was the constraint. Building on a phone in the early days forced smaller scope, tighter asks, cleaner output, because you physically can't sprawl on a six-inch screen. That habit never left, even once I moved to a laptop.

The other thing nobody tells you: the hardest products weren't the ones with the most moving parts. They were the two where I had zero prior foundation to stand on — the model fine-tuning and the Android app. Everything else extended a skill I already half-had. Those two didn't.

I remember genuinely asking myself "is this even mine to attempt" before starting either one. Both got done.

I'm not writing this to tell you to go build twenty-six things in your spare time — I'm not sure that's a healthy template for anyone, myself included some weeks. I'm writing it because I think there's a real difference between reading about a technology and actually building something with it until it breaks, and almost nobody talks about how uncomfortable that second part is.

If you want a governance framework, an opinion on whether your team should use a given AI tool, or just want to know if the thing you read about actually works — I've usually already broken it myself, on a Tuesday night, trying to make it behave.