Why should you believe me?

The short answer is that you shouldn’t. Believe the output instead. I’d much rather be humble in how I talk about things and let the work speak for me.

I’ve read countless articles in the past few months from people who confidently claim to know what’s going on in this industry now that AI is playing a big part in it. Most of them don’t. I’ve also read many articles from people who really do know what they’re talking about. Telling the difference between the two is legitimately difficult, and I won’t claim to be one of the people who knows. What I’d rather do is show you my work: my mistakes, the lessons I’ve learned, the failure modes that happened, the best practices I’ve discovered, the methodologies that mattered, and what the table stakes are when you’re dealing with agents instead of people.

I do think you’ll find there are a lot of parallels between the two. I certainly have, and maybe more similarities than differences when you get right down to it. There will be a whole note on that.

I’m learning this alongside everybody else, and I’m going to try not to make too many predictions. The industry is changing too fast. Models are released weekly, upgrades land constantly, and tooling that didn’t exist a month ago is now considered the standard. Keeping track of it all is hard. So what I’m going to talk about instead is the things I’ve done and what I learned doing them, in the hope that those lessons hold regardless of the tooling.

This isn’t a course. These articles are online and not behind a paywall, and won’t ever be. I’m hoping this has some value for people who are on the same journey I am.

Here’s what I will claim. I’m shipping code, in this case a game. It’s production-grade work, not a hobby. I consider it my full-time job at the moment (thanks to my lovely wife for giving me that freedom), and I put in as many hours a week as any job I’ve held, and often more. I’m just doing it with a team of agents instead of a team of people. And I’ll say this: it’s the most fun I’ve had in a long time.

I try to keep things organized. The tabs on the edge of the page sort the notes into what I think are reasonable buckets. The FAQ tab is answers to questions people have actually asked me. There’s no schedule. I have a lot of content I’ve already thought through, so there’s plenty to say, and as those thoughts turn into finished notes they’ll show up here.

If you don’t find this helpful, if you don’t like the articles, if the headlines don’t catch your attention, just stop reading. No harm done, and I won’t take more of your time.

There’s no secret formula in here. It’s a logbook: what worked, what didn’t and what changed along the way.