I Handed My App Studio to AI #3 · October 24, 2026

AI pitched 24 app ideas. The owner rejected all 24. Here's what it learned.

We asked AI (Claude) for our next app idea. Over three rounds it pitched 24, and the owner said no to every one. The reasons attached to the last four, and the conclusion the AI drew.

Key takeaways

At hiio, a one-person app studio, AI (Claude) pitched 24 ideas for the next app over three rounds (10, 10 and 4), and the owner rejected all 24. The reasons were that the scenes felt trivial, forced or gimmicky, already seen, or not to the owner's taste. The AI concluded that ideas assembled at a desk don't work and should start from something many people already do, and it switched from building a new app to improving the existing app, Outgo.

We asked AI (Claude) for our next app idea. It pitched 24 over three rounds, and the owner's answer to all 24 was no. The short version: the ideas looked well-formed, but none started from something people already do, so they all failed. Instead of a new app, we're now improving the one we have.

Previous part: 14,500 views, 3 downloads — week 3 report card for an app run by AI

24 ideas, by the numbers

Idea names below are translated from the Korean working titles.

Round	Pitched	Result	Examples
1	10	All rejected	Our Own Radio, Photo Album Docent, Step Mail, Group Chat Newspaper
2	10	All rejected	Payday Meeting Minutes, Couple Diary for One Year Later, Streaming Sub Rotation
3	4	All rejected	Plans Billboard, Group Chat Election Night, Blind Music, Taste Stocks
Total	24	0 approved

The owner answered each idea with a yes or a no. That's the studio's rule: the owner only signs off.

The reasons on the last four

The four ideas in round 3 came back with reasons:

The scene is trivial
Feels forced, like a gimmick
Seen it before
Not my taste

What the AI concluded

Ideas assembled at a desk around a format don't work. Rather than ideas that invent a new behavior ("it'd be fun if friends did X together"), start from something lots of people are already doing.

So new apps are paused for now, and the work has moved to improving Outgo, which is already on the App Store. New ideas no longer go straight to building: they go through a candidate list, scoring and a small validation test, and get built only after approval.

FAQ

Does AI come up with usable app ideas?

In this experiment, 0 of 24 were approved. The ideas came quickly and were well-formed, but the rejections were mostly "trivial, forced, seen it." Without real behavior people already show as the starting point, the ideas tended to stay plausible combinations.

What happened after all 24 were rejected?

Building a new app stopped, and the work shifted to improving the already-released app, Outgo. Future ideas now go through a candidate list, scoring and a small validation test before the owner approves them.

How did the owner respond to each idea?

With a single yes or no per idea. Only the last four came with short reasons.

Sources

All episodes of 'I Handed My App Studio to AI'

#AI ideas#Claude Code#app ideas#indie dev#build in public

Written with AI (Claude) and checked against the linked sources.