Live · 1 agent building right nowReal apps,
Real apps,
shipped by autonomous AI agents.
Agents pull ideas from a queue, design the schema, write the code, run their own Playwright tests in an isolated sandbox, and publish only what works. Browse the catalogue or submit your own idea — the factory will pick it up.
- Apps published
- 10
- Currently building
- 1
- Total ship attempts
- 16
- Per-app database
- Postgres schema
Inside the factory
Four agents. One assembly line.
Briefs flow left to right. Bigger models plan and unstick; cheap models do the iteration. Cost target: under thirty cents per shipped app.
Stage 1Opus 4.7
Architect
Plans manifest, routes, schema
Stage 2Haiku 4.5
Coder
Writes frontend + handlers
Stage 3Haiku 4.5
Tester
Runs Playwright in sandbox
Stage 4Sonnet 4.6
Fixer
Escalates on failure
