i.
Vision-AI auto-tagging
One photo. Gemini Vision returns category, colour family, pattern, material, season, formality, and a tasteful style descriptor — all structured, all editable.
A Happier Style product — Issue No. 01
AI Closet attacks a small, daily kind of friction — the mental load of deciding what to wear. Photograph your clothes once; let the app tag, organise, and recommend outfits for the weather, your calendar, and the way you want to feel today.
How it works
AI Closet does the patient, unglamorous work of remembering. The taste, the styling, the verdict in the mirror — those stay yours.
01
Batch-photograph your wardrobe with the in-app camera. AI Closet segments each garment, then sends a single, high-quality image through vision AI to extract structured metadata — category, colour, fabric, season, formality. You do this once. The closet remembers forever.
02
Recommendations don't re-process pixels every time. AI Closet reasons over the lightweight text metadata it already extracted — fast, cheap, and good. Ask "wedding outfit, 50°F, looks polished but comfortable" and it composes from what you own.
Pl. 03 — Query in, outfit out. 03
Tap the look you actually wore. Over time, AI Closet learns the way you dress — the silhouettes you reach for, the pieces that sit unused. A quiet OOTD record, not a streak; a wardrobe audit, not a leaderboard.
Pl. 04 — The OOTD log, gentle by design. Features
i.
One photo. Gemini Vision returns category, colour family, pattern, material, season, formality, and a tasteful style descriptor — all structured, all editable.
ii.
Today's forecast, the events on your calendar, the temperature you actually feel cold at. The model reads the room and dresses you for it.
iii.
Quick — three to five outfits, in under a second. Exploration — a deep dive when you want to surprise yourself with something you forgot you owned.
iv.
Real-world constraints, built in. One pair of shoes per outfit. No two tops. Layering rules respected. The model can't recommend nonsense because the schema won't let it.
v.
Your closet lives in a local SQLite cache (Drift) on-device. Open the app on an airplane, on the subway, anywhere — it just works. Cloud sync (Firestore) is opt-in.
vi.
A daily record of what you actually wore. Useful for the AI; quietly useful for you, when you wonder what worked at last year's wedding.
Mission
02
Most personal-styling apps treat your closet as a feed to scroll. AI Closet treats it as a database to reason over — the difference between a magazine and a librarian who has read every issue.
The heavy AI work — segmenting an image, extracting attributes, checking colour theory — happens once, at the moment a garment first enters your closet. From then on, recommendations are fast and cheap because the model only ever reads structured text. Your photos sit locally; the model reasons over words.
The end state we're after isn't "more outfits per minute." It's a measurable rise in the style savviness of an entire generation — fewer panicked mornings, more confidence in clothes you already own, less waste.
FAQ
Get the app
We're hand-rolling invitations while we polish the rough edges. Drop your name and we'll write back personally — no marketing list, no "thanks for subscribing" sequence, just a real email.
iOS — TestFlight, summer. · Android — internal track, autumn.