Curate

Inspiration

We recognized that many people have clothes that they enjoy wearing, but they aren’t necessarily highly knowledgeable in fashion. A personal stylist should not be limited to just high-end clients and celebrities. With one of our founders having owned a clothing brand, he has seen firsthand how customers crave outfit inspiration and styling advice, and Curate provides exactly that.

What it does

Curate helps you find your style. First, users scan their wardrobe, then our stylist learns what they own and identifies their style preferences, and offers potential style paths the user could potentially go down. It can build outfits from existing clothes, spot gaps in their wardrobe, and recommend purchases that would unlock more outfit combinations.

How we built it

We built a TypeScript and Node.js backend connected to Relay for chat and calls. Gemini analyzes wardrobe videos, plans outfits, and generates personalized previews. The Stylist passes clothing criteria to the Shopper, which uses Google Search grounding to find real retailer listings. Supabase stores wardrobes, preferences, and message history, while ElevenLabs provides the voice integration.

Challenges we ran into

Making voice calls reliable and handling slow product searches were major challenges. We also had to make sure the assistant distinguished clothes users actually own from items they might want to buy.

Accomplishments that we're proud of

We connected wardrobe videos, three distinct style pathways, personalized outfit previews, and real product recommendations in one conversation. The assistant also learns from feedback and remembers different spending limits for different clothing categories.

What we learned

Personalization requires listening to why someone dislikes a suggestion, not just recording a rejection. We also learned how to divide work between agents, collaborate through GitHub, and test the complete user experience.

What's next for Curate

We want to transform Curate from a project to a long-term personal style platform. In order to do so, we would implement a function that allows a subagent to purchase clothing items for the user after approval. The user experience and personability can be enhanced through functions such as event-specific styling, weather-aware outfits, price-drop tracking, and the virality and shareability can be amplified through wardrobe rating, weekly recap, wardrobe personality features.