I want to go back to the very start of this — the first time I opened a chat and described an app that didn't exist yet — because the most common question I get isn't about the app. It's some version of: "Wait, you built that? But you're not a developer."
Correct. I'm not. I've spent years close to technology — products, systems, the language of how software gets made — but I don't write code for a living, and left to my own devices I couldn't have shipped this. So the honest answer to "how did you build it" is: I didn't, alone. I built it with AI, as a genuine working partner. And the interesting part isn't that AI exists — it's how the work actually divides up between a non-developer and a machine when you're trying to make something real.
What I brought, what it brought
Here's the split, as honestly as I can put it.
I owned the decisions. What the app should be. Who it's for. Why one companion app across every kind of live event instead of yet another single-category tool. Which features matter and which are distractions. Whether to build a social network or route around one entirely. How to think about the thing that makes it defensible. None of that came from AI — and when I asked it to weigh in, the useful answer was always "here are the trade-offs," never "do this." The judgment was mine to make, and it stayed mine.
AI owned the execution I couldn't do. Turning "I want users to be able to save photos from events they've been to" into actual working code, with the right database structure, the security rules, the edge cases I wouldn't have known existed. The thousand small technical decisions underneath a feature that I can specify but couldn't implement.
The mental model that clicked for me: it's less like having a tool and more like having a very fast, very knowledgeable collaborator who never gets tired — but who needs me to know what we're building and why. It doesn't replace the thinking. It removes the wall between the thinking and the thing.
Where it's genuinely hard
I don't want to oversell this, because the hype usually does, and that helps no one.
AI is brilliant at the code and surprisingly weak at the context. It'll confidently build exactly what you asked for — including when what you asked for is subtly wrong, or when the right move was to push back on the request itself. The skill I've had to develop isn't coding; it's knowing what to ask for, recognising when an answer is off, and holding the overall shape of the thing in my head so the pieces actually fit together.
And a lot of the real work isn't the building at all. It's the unglamorous stuff around it — the environment that won't cooperate, the deploy that silently fails, the bug that takes three hours and turns out to be one wrong word. AI helps enormously with all of it, but you're still the one steering, deciding, and deciding again.
Why I think this matters
If you're technical-adjacent — you get systems, you have ideas, you've just never been the person who writes the code — I think the gap between "I have an idea" and "I have a working thing" has quietly collapsed. Not to zero. You still need judgment, persistence, and a willingness to sit in the confusion when something breaks. But the part that used to require either years of learning or a budget to hire it out is now something you can genuinely collaborate your way through.
That's what this journal is really documenting, underneath the app: what it looks like when someone who isn't a developer builds real software anyway, with AI doing the half I can't and me doing the half it can't.
More soon — including the parts that went wrong.