All Build Log posts
Build Log

I can't code. I shipped 3 real apps anyway.

I did not come from a developer background. I still shipped three real apps with AI. Here is what was true, what was hard, and what actually made it work.


The short version

  • I did not study computer science, work as a developer, or learn frameworks.
  • Three real apps are live anyway — with real interfaces, real user flows, and real bugs fixed.
  • AI made software buildable through direction, not traditional coding fluency.
  • What changed was not that AI became perfect — it did not. What changed was how I worked with it.

The mindset shift that made everything else possible

  • I stopped trying to become a developer before building.
  • I became a better director instead — someone who describes clearly, reviews carefully, and decides what good looks like.
  • The line that changed how I worked: one change at a time, every single time.
  • Asking AI to do too much at once is the most common reason non-developers stall and give up.

What the director workflow actually looks like

  • Break the product into one-screen or one-problem chunks — never ask for the whole thing.
  • Describe exactly what already exists before asking for a change.
  • Explicitly tell AI what not to touch, not just what to change.
  • Check every result before moving to the next task — fast without review is fake speed.
  • Save a working version to GitHub before every change, not after.
  • Treat broken screens as normal signals, not as evidence you should quit.

What I got wrong early on

  • Asking for giant rewrites instead of one targeted change.
  • Trusting AI output without reviewing it properly — changes I had not checked broke working screens.
  • Thinking one more quick change would be fine — it almost never was.
  • Losing confidence the moment something broke instead of treating it as a normal development step.

The question that reframes everything

  • Could you have built a live landing page a month ago? If no — progress is happening.
  • Could you have shipped a simple app flow six months ago? If no — keep going.
  • Do not compare yourself to engineers with ten years of context. Compare yourself to last month.
  • The goal shifts from "can I build an app?" to "what is the next smallest thing to build?" — that is the whole game.

Why NonDev Apps exists

  • Not to promise overnight results or cosplay startup guru energy.
  • Not to pretend there are no frustrating days — there are.
  • To show non-developers that real shipped work is possible with a safer workflow and enough persistence.
  • Start with the free 5-Day Validation Landing Page challenge if you want the simplest entry point.
I can't code. I shipped 3 real apps anyway. — NonDev Apps