My first AI experiment is finished: an app from idea to production, with me as the only human on the team! Before I jump into the next experiment, I want to share what I'm actually taking with me.
First, a confession, because it colours everything below: Sure, I can't write code… but… I have twenty years of experience commissioning, specifying and leading systems development, and that experience turned out to be worth more than I'd thought.
Yes, everyone has said this already, so I'll keep it short: It's true! In a few weeks I learned to work in Cursor (really like it!), Supabase and Vercel with Claude as the engine, and went from zero to a live app. The door that used to be shut to us non-developers is now firmly ajar.
When building apps, sites and systems cost a lot, the cost worked as a natural filter against bad ideas. That is, you thought before you acted. When building becomes almost free, that filter sadly disappears, and the world will fill up with well-built apps that nobody needs. So ideation isn't just important any more — it's the new bottleneck. Think the idea and the design through carefully before you start building. And here AI is a very good companion.
I noticed it myself: the biggest benefit of AI didn't come in the coding but before it. As a sounding board that questioned and refined, and that on a few occasions made me cut features I'd wanted but that turned out not to be so well thought through.
Things are moving furiously fast and new tools and AI services show up every week. It's easy to dive down a rabbit hole and come up three evenings later without having created any value. The most important discipline turned out to be entirely non-technical: keeping my eye the whole time on what I wanted the AI to help with, and not getting stuck in tech-tinkering for its own sake.
What surprised me most wasn't the technology but the comment threads and the face-to-face conversations: "I've built an app too that…", "I've been trying this AI thing…". And it isn't just developers — just as often it's non-technical people. It's almost as if there's a grassroots movement of people tinkering at their kitchen tables. Some organisations probably underestimate how many of their own people are already up and running at home.
My next experiment starts from a simple question: What can AI really do, in practice?
More on that soon. And as always: do tell me I'm wrong in the comments — that's how the lessons get better.
I share what I build, test and learn as I go — both here and on LinkedIn.
