I had the idea for BaseFuel in January. By the end of February I had something running on my own phone. This is the workflow that got it there.
Start from the friction
Every other nutrition app I tried treated logging as the product. The actual product I needed was knowing whether today’s training was supported by today’s eating.
So I wrote the smallest spec I could that answered that question and nothing else. No social features, no gamification, no streaks.
Let Claude Code do the boilerplate
SwiftUI scaffolding, HealthKit wiring, Core Data persistence — all of it generated and iterated in a tight loop. I’d describe the behaviour I wanted, review the output, push back on anything that felt over-engineered.
The interesting work was in deciding what not to build. Claude Code is very willing to add features. You have to be the one who says no.
Ship to one user first
That user is me. The TestFlight build went to my own phone for three weeks before anyone else saw it.
By the time I opened it up, the obvious bugs were gone and the shape of the thing was clear. People knew what they were getting.
What I’d do differently
I’d write the data model earlier. I changed the schema twice mid-build and it cost me a week each time. Next project starts with the model, not the UI.