Newsletter

It wasn't AI doing its best job. It was me doing none of mine

This year I shipped something I'd been putting off for most of a decade: my own iOS app. After 9+ years of writing Android for a living, I built a journal app that asks you exactly one question a day; in SwiftUI, on weekends, alone.

Except "built" deserves an asterisk. I didn't write a single line of code. I orchestrated the AI agents that did. Work I would have budgeted a month for was done in a weekend, and I remember feeling like I'd gotten away with something.

Then, over the following weeks, I watched three people in my circle ship journaling apps of their own. More or less the same features as mine. More or less the same design as mine. Nobody had copied anybody.

That's not a coincidence. That's a supply chain.

The audit

When I traced back how my app had actually come to exist, the pattern was embarrassing:

I hadn't come up with the idea, I'd asked AI what I could build in a weekend, and it suggested what it suggests to everyone: a journal, a weather app, a to-do list.

I hadn't designed it, I'd told AI to apply "UI/UX best practices," and it produced the same tasteful-but-anonymous design it produces for everyone.

I hadn't chosen the features, I'd asked what features a journal app should have, and it gave me the union of every journal app that already exists.

I hadn't even thought about how to reach users, I'd asked for the best ways to promote an app, and received the same launch checklist as every other weekend builder.

It wasn't AI doing its best job. It was me doing none of mine.

That's what AI slop actually is. It's not a property of the tool. Every decision you hand over completely, the model fills with the statistical average of everyone else's decisions. Hand over all of them and the output is the average, and now you're holding a perfectly generic app in one of the most crowded categories on the App Store, wondering why it doesn't stand out.

The fix isn't less AI

It's more you.

Everyone has one area of unfair advantage. Maybe you generate product ideas nobody else has. Maybe you have a designer's eye. Maybe you have unusually good instincts for which features matter, or a knack for finding the people a product is for. The move is to identify that one area, and refuse to delegate it. You pour yourself into that layer, and let AI do everything downstream of it.

If you're a designer, you put design ideas into AI, never the other way around.

Mine is design. It's literally been my job. So since version 1.0, that's how One Question gets built.

The calendar became a field of dots, a whole month readable at a glance, nothing shouting at you. Navigation became a swipe: screens sit side by side like pages in a notebook instead of hiding behind menus. Everything nonessential got cut, then cut again. And the detail I'm proudest of: every daily question ships with its own placeholder text, written specifically for the moment your brain goes blank. Because the hardest part of journaling was never the writing, it's the starting, and a generic "Type here…" has never helped anyone start.

Every screen decided by me, built by the machine. It's still a work in progress, but the app finally has taste it didn't have at launch, and I can tell you exactly which parts of it couldn't have come from anyone else, because for a while, all of it could have.

The quiet result

There are now people in Sweden, India, United States, United Kingdom, Austria, South Africa, South Korea who open One Question every day and write. Over 170 entries between them. By industry standards those are rounding-error numbers. But they're strangers, they formed a daily habit around something I made, and they keep coming back.

Building is no longer the moat. When an app takes a weekend, everyone's weekend produces one. What's left is the part that was always the point: knowing what's uniquely yours, and refusing to let the tool average it away.

This is the first lesson from building One Question in public. There will be more, building, designing, and distributing an app solo, in an industry that won't hold still, generates them faster than I can write them down.

One Question is on the App Store. If you try it, I genuinely want the honest version of what you think. 👇