Tryzeon

Tryzeon Fashion Agent is an AI stylist any fashion brand can embed with one line of script. Shoppers say where they're going, or show a photo they like, and the agent builds a complete outfit from the brands' real stock, checks every piece against their body measurements, and renders the whole outfit on their own photo. It then inspects its own render against each product photo, fixes what doesn't match, and turns the result into a shoppable image: tap the jacket, open the product. The problem: shopping is moving from search to AI agents, but agents can only recommend what they can read. Small fashion brands' catalogs aren't AI-ready, and general assistants don't know what's in stock, what fits you, or how it looks on you. Tryzeon is a Taiwanese startup already live on LINE with partner brands; for this hackathon we rebuilt our recommend-to-try-on loop from scratch on Meta's models as an embeddable web agent. Results on our 10-request benchmark: 10/10 outfits valid, in stock and fitting the shopper's measurements; try-on fidelity passes rose from 6/10 on the first render to 10/10 after the agent's self-check and fix loop. Live demo: https://tryzeon-fashion-agent.vercel.app