How long does it take to build an app with AI?

Weeks of evenings, not years — but not a weekend either. Working two hours on weeknights plus Saturday mornings, a real product takes me a few weeks to reach launch, and longer to reach the version I'm not embarrassed by. Anyone promising a serious app in a day is selling something. Here's where the time goes.

The demo comes fast

The first working version is genuinely quick now. A screen, a flow, data being saved — with vibe coding that can exist within the first few sessions. This is the part the hype videos show, and the part that makes "I built an app in 24 hours" technically true and practically misleading: nobody's data is in a demo.

The last 20 percent is most of the calendar

What stands between demo and product, launch after launch: accounts and login. Payments. The empty states nobody plans for. What happens when the network drops mid-save. What happens when a user does the thing you never imagined.

If you're shipping to phones, add store review. PairHabit's App Store release was its own education — guidelines, review cycles, launch mechanics I learned live, on the clock.

Scope moves the number. Skill doesn't.

Scope, more than skill. A single-purpose web tool is weeks. A two-sided app that syncs between people is months of margins. Platform matters too: web ships in one step; iOS adds Apple to your schedule.

And your hours. My math is simple: two hours a night, five nights, plus a Saturday morning is around 13 hours a week. So a "three month" project is really a 150-hour project. Count your hours, not your months — it's the only estimate that survives contact with a full-time job.

Where that leaves you

The tools removed the years of training. They didn't remove the work. And if you're building around a job and a family, the schedule itself becomes the skill — that one gets its own field note.