Inspiration
My inspiration for Ensemble AI came from a universal, daily struggle: standing in front of a closet full of clothes and feeling like you have "nothing to wear." We saw an opportunity to use the power of modern AI to solve this problem, not just for individuals, but as a scalable enterprise solution.
Our vision was to create a B2B2C platform that could be offered by any fashion retailer. By empowering their customers with a personal AI stylist, retailers can drive deep engagement, increase customer loyalty, and provide a truly personalized shopping experience that leverages their own product catalog. We wanted to build an agent that doesn't just show you what you own, but truly understands your personal style to make getting dressed effortless and fun.
What it does
Ensemble AI is a fully interactive web application that serves as a complete fashion assistant with three core intelligent functions:
Digital Wardrobe: Users can upload an image of any clothing item from their closet. Our multimodal AI (powered by Google Gemini) analyzes the photo in real-time, extracting key objective attributes like its type ("T-Shirt"), color ("black and white"), and style ("casual, striped"). This structured data is then added to the user's personal digital wardrobe.
AI Stylist: The user can request an outfit for any occasion, from a "Sunday Brunch" to a "Work Meeting." The agent intelligently combines items from their personal wardrobe, generating a complete, context-aware outfit recommendation and explaining the styling choices, just like a real stylist would.
Personal Shopper: The agent analyzes the user's unique style based on the contents of their wardrobe. It then searches a comprehensive, built-in product catalog to recommend new items to buy that will perfectly complement what they already own, creating a powerful upselling and cross-selling opportunity for a retail partner.
How we built it
• Python for the backend logic
• AWS Bedrock for multimodal AI (Claude 3 Haiku)
• MongoDB to store wardrobe and product catalog data
• Bright Data to store wardrobe and product catalog data
Log in or sign up for Devpost to join the conversation.