Thread
AI wardrobe and styling app · iPhone · Swift, SwiftUI, Supabase · 2026 to present · In development
The question
I own clothes I forget about, and I wear the same few combinations. Most wardrobe apps want me to photograph every item against a blank wall first. Could an app build a digital closet from the photos I already have, so the setup feels as easy as uploading a picture?
What it does
You upload an ordinary photo: a mirror selfie, a vacation shot, a group picture. Thread finds the clothing and accessories in it, pulls them out, and can turn them into clean, catalog-style product images. Those become a digital wardrobe you can browse like a fashion catalog.
- Finds several items in one photo, including tops, bottoms, shoes, and accessories
- Reconstructs clean product images while keeping the real color, shape, and details
- Works with clothes that are partly hidden, and asks for a close-up when the evidence is thin
- Tries to recognize items you already saved, so the same shirt does not get added twice
- Suggests outfits from what you own, and factors in the weather through Apple WeatherKit
How it is built
A native SwiftUI iPhone app. Supabase handles authentication, data, and edge functions. Computer vision and AI models handle garment recognition, segmentation, and the catalog image reconstruction. Weather comes from Apple WeatherKit. The app sends images to AI providers for recognition and reconstruction, so it is not fully on-device.
Where it stands
In development, not on the App Store. The core pipeline runs today: multi-item recognition, extraction, catalog image generation for eligible items, basic outfit and styling help, wear logging, and weather in the app. Reconstructions that are uncertain ask me to confirm them, rather than passing a guess off as the real item.
Still being validated: accessory image fidelity on real devices, matching the same item across different photos, and some harder group-photo and hidden-clothing cases.