AI personal stylist
ClosetPilot
An AI styling assistant: photograph your wardrobe once, then get outfits for any event, a short list of what is missing, and brutally honest feedback on the look.
- Role
- Design and build, end to end.
- Status
- In development
- Year
- 2026
Private app running on the owner's own computer. Wardrobe photos are blurred except the ones approved for this page.
The problem
Most style advice either sells you new clothes or flatters you. Neither helps you get dressed from the wardrobe you already own, for the evening you actually have.
What I did
Designed and built the whole app: wardrobe import with evidence, the outfit library, an event-aware stylist, a budget-capped shopping list, and an honest mirror for outfit reviews.
How it works
Photos are imported once, checksummed and de-duplicated, and every garment keeps the photo evidence it came from; it stays "observed" until you confirm you own it. The stylist ranks real outfit recipes by occasion and style lane, shopping only fills genuine gaps inside a hard budget, and the honest mirror judges the outfit, never the body, with fixes ordered by impact. Everything runs and stays on your own computer.
Captures
The hard parts
- 01
Never inventing a wardrobe: a look with a missing piece cannot be logged as worn until you link something you really own.
- 02
Honest without being cruel: the mirror scores the outfit and the fit, never the body, and every critique comes with a specific fix.
- 03
Private by design: photos, thumbnails and the database stay local, with no account, no analytics and no auto-purchases.
What changed
- A working local app: 21 wardrobe pieces and six looks imported, 23 unit tests and five browser test suites passing.
- Next: a live vision model, so new photos get honest scores instead of reference observations.
Built with
- Next.js
- React
- TypeScript
- Tailwind CSS
- SQLite
- Sharp
- Playwright