Inspiration
Most people who start learning a language quit. Not because it's hard. Because it's boring.
It's 11pm. You've got 40 minutes. You open the "Spanish" row on Netflix, scroll past a hundred things you'd never watch, and put on something in English. Again. That's where the dream dies.
The problem was never finding Spanish movies. It's finding one you actually want to watch.
What it does
You tell Homegrown three movies you love. You pick the language you're learning. That's it.
Homegrown shows you the movies you would have loved if you'd grown up there. Real local films, picked for your taste. Each one tells you why it's yours, and whether it's easy, medium or hard to start with.
Then you get a postcard to send a friend: "My film self grew up in Seoul. Who's yours?"
26 cities. Seoul to São Paulo, Tokyo to Cairo, and India state by state.
How we built it
Qloo knows taste. So we let Qloo pick. Then we checked every single film against TMDB to make sure it's really from there, in the right language. If it isn't, it doesn't show up. No filler. Ever.
Where Qloo doesn't have enough films yet, like Kannada or Marathi, Claude suggests some. We verify each one, and we tell you plainly that the AI picked it.
Vanilla JS, Node on Vercel, 270 tests, every push scanned for secrets.
Challenges we ran into
Saying "local" is easy. Proving it is hard. So we measured. Music, TV and movies, across 40+ cities. A city only goes live if most of its picks are genuinely from there. Movies won: 93% local. So we shipped movies first, and said no to everything else.
What we learned
The most important thing we built was a number that told us where not to launch. Focus is saying no.
What's next
Where to watch each film. One perfect first movie for beginners. And seeing how your postcard compares with your friend's.
Built With
- anthropic
- claude
- css
- github-actions
- html
- javascript
- node.js
- qloo
- serverless-functions
- tmdb
- vercel
