Inspiration

Closer.ai started from a simple frustration: inspiration is everywhere, but turning a vibe into an actual outfit is slow and unclear. We wanted to build something that could take extremely simple prompts like "JFK Jr. spring" or "Rihanna x Zendaya" and instantly translate that into real, shoppable looks.

What it does

Closer.ai turns style prompts into complete, purchasable outfits. It understands and parses the "vibe," converts it into structured style DNA, scraping prices, URLs, and images to find real products, rank them by fit and price, and explains why each piece works. Users can refine results by budget or occasion, and even upload a photo to get matched with a style icon.

How we built it

We combined LLM based reasoning (standard text output to understand/go more in depth about the style the user wants) with a backend that structures style into attributes, then connects to live product and pricing data. A ranking system scores items on relevance and practicality, while APIs handle retrieval, processing, and secure rate limited delivery.

Challenges we ran into

Getting/scraping consistent, high quality product data (especially images and pricing) was difficult. Another challenge was making outputs feel truly personalized instead of generic, while keeping response times fast.

Accomplishments that we're proud of

We are primarily proud that we built a full system that goes from a vague idea to a fully shoppable outfit in seconds, with reasoning behind every single recommendation. The lookalike, and trends features and some refinement make it feel way more interactive.

What we learned

We learned how to turn subjective concepts like "style" into data, integrate LLMs with realtime systems, and design ranking algorithms that balance taste, cost, and usability.

What's next for Closer.ai

We plan to improve recommendation accuracy, expand product coverage, and add deeper personalization. Long term, we want to evolve Closer.ai into a full AI stylist with social and premium features.

Share this project:

Updates