# Autumn's Second Bloom

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Autumn Thomson](https://cerebralvalley.ai/u/VerdantForged)
- **GitHub:** https://github.com/codepilots/meta-styles
- **Demo video:** https://youtu.be/RGsLoOL9Gx4
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3452

The problem

Most wardrobes now live in photos — new buys, old favourites, selfies — but photos never become outfits. Styling apps start from a shop's catalogue of products rather than the wardrobe you actually own, so their suggestions answer someone else's closet, not yours. Generic AI assistants compound that: they propose combinations that break the basic rules of dressing — a base garment with nothing finishing it, layers doubled up — invent products and prices they have no data for, and finish with words when what you need is to see the look.

What meta-styles does

meta-styles is an Instagram-DM-style chat where your photos become a persistent wardrobe that answers back. Share a photo and SAM 3.1 segments it into garments; Muse Spark 1.3 classifies each piece and assembles full looks from your own wardrobe; every proposal passes a real fit-rule engine (base + exactly one finish; two-layer and shoe rejections) before it is sent. Ask for a preview and Muse Image 1.0 renders the assembled look on you — the payoff is the picture, not a paragraph. Say "no bag" and the look is revised and re-validated. The loop closes with one purchase suggestion that genuinely completes the look, and the agent is honest: without fabric or price data it says so rather than inventing a listing. That loop serves a disorganized wardrobe at home — and the same photo-to-validated-look loop, gives retailers a way to style complete looks from a customer's own pieces.

## More from Meta Global AI Developer Hackathon

- [NailMentor by IND Academy](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3449)
- [DORA](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3450)
- [Nanjo](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3451)
- [AVO](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3453)
- [TraceForm](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3454)
- [ASCII](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3455)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3452. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt