I couldn't get my daughter to clean her room, so I built an app where AI grades the before/after

Hey IH. Solo dev, first project I've actually shipped, so I'm a little nervous posting this.

It started with my daughter's room. I tried asking nicely, a chore chart, small rewards, negotiating, losing my temper. Nothing survived more than a few days. The one thing that clearly does motivate her is what her games do: XP. A number that goes up, and something outside her confirming she earned it. So I built her a referee for cleaning.

You photograph the mess, clean, photograph the result. Claude Vision compares the two shots, scores the improvement 1 to 10, awards XP and tells you what's still wrong. Around that: levels, streaks, 19 achievements, a leaderboard, and a feed where friends react to each other's before/after.

Then it started working on me, which I hadn't planned for. A todo item that says "clean the kitchen" sits there forever. A 6/10 with a note that the counter is still cluttered actually gets me up to finish it.

Some things I got wrong along the way.

The hard part wasn't the AI, it was making the app feel calm. Pages flickered on every navigation and it took me embarrassingly long to see why: the platform's own slide animation and my own fade were fighting each other, so a page arrived fully visible, dropped to zero opacity for a frame, then faded back in. That whole class of bug is invisible in code review and obvious the second you hold the phone. Stack, for anyone curious: .NET MAUI for one Android and iOS codebase, a C# backend on Railway, Supabase, Claude Vision for the judging.

Streaks I got wrong twice. A daily cleaning streak dies around day three and a dead one demotivates worse than none, so the main streak is moving to weekly. The code side is still open: production showed a streak of 1 for a user with no cleanings behind it, a local time versus UTC mess I patched without being sure I found the cause.

The thing I'm worst at is distribution, and I went in with no audience and no experience. My first plan was Polish Facebook groups about cleaning and home organisation, since people there are already trying to make themselves tidy up. Then I joined and read what actually gets posted: mostly cleaning companies advertising and people asking how to get a stain out, not communities that want a motivation app. My best channel went to almost zero before I posted a word. I expect to be wrong about the next channel too, just faster.

The app is going into closed beta now: Google Play on Android, TestFlight on iOS. One caveat on timing, because Google's rule shapes it. A closed test needs the group to start together and stay opted in for 14 straight days, so Android invites go out in one wave in early August rather than the minute you sign up. iOS is immediate.

The first 50 testers who stay active through the 14 days get Premium for life and a Founding Tester badge in the app. It's free, I'm not selling anything, I just need real people with real mess.

https://ordoapp.net/beta?source=indiehackers

And if you've found your first users with no audience to start from, or run an app through Play's closed testing, I'd take any advice.