🛍️ ShopTheLook

🚀 Inspiration

Ever stared into your closet and thought, “I don’t even know what vibe I want today, let alone how to dress for it”? We felt that too.
Finding clothes that match how you feel—then figuring out what to wear, what design fits, and where to buy it—is a whole process.

So we built ShopTheLook: an AI-powered fashion assistant that removes the hassle of styling and shopping. Whether it's a mood, a vibe, or just a random cosmic thought, ShopTheLook translates it into ready-to-wear fashion suggestions you can shop instantly—powered by multimodal AI Gemini.

👗 What it does

ShopTheLook frees you from:

❌ Brainstorming what to wear
❌ Googling how to describe your mood in fashion terms
❌ Searching across dozens of sites to find matching outfits

Instead, it uses the full power of Gemini's multimodal intelligence to do it for you:

🧠 Text Input → Refined Prompt: Gemini takes your raw thought and transforms it into a creative, fashion-forward prompt perfect for image generation
🎨 Prompt → Image: Stable Diffusion creates high-quality, unique outfit visuals based on that refined idea
👁️ Image → Search Phrase: Gemini's vision model kicks in, analyzing the AI-generated image and crafting a clean, search-optimized phrase
🛍️ Search Phrase → Product: SerpAPI searches Google Shopping to bring you real products that match the vibe
⚡ All of it gets cached with Pinecone for instant recall and performance boost

Whether you're going for “sunset overdrive”, “celestial royalty”, or just “cozy coffee run energy”, ShopTheLook makes it real in seconds.

🛠️ How we built it

Gemini API (Multimodal):
- 🔡 Text mode: Refines vague user prompts into clear, model-ready fashion concepts
- 🖼️ Vision mode: Analyzes generated images to produce realistic and shoppable search terms
Stable Diffusion (via HuggingFace): Generates the outfit visuals
SerpAPI: Pulls real-time Google Shopping product listings
OpenAI Embeddings + Pinecone: Caches prompt-to-result connections and boosts response speed
Google Cloud Storage: Hosts and serves generated images
LangChain: Orchestrates prompt flow and response pipelines
Google Colab: Used for building and running the system

😤 Challenges we ran into

Gemini occasionally generated prompts that were too long for Stable Diffusion's token limit
Ensuring gender-accurate styles from generative outputs
Cleaning and formatting image analyses into product-searchable terms
Speed optimization: balancing visual quality and API latency
Managing API keys and secrets across several cloud tools

🏆 Accomplishments that we're proud of

🔁 A complete AI fashion loop: Mood → Prompt → Image → Description → Shopping
🌟 Gemini acting as a dual-purpose brain, handling both creative prompt generation and visual understanding
⚡ Pinecone-powered memory for sub-second performance on repeated prompts
🎯 Created stunning, imaginative outfit designs based on abstract and emotional inputs

📚 What we learned

Multimodal AI like Gemini changes the game when it comes to creating meaningful, user-friendly experiences
Prompt engineering + vision models = magic
Caching with Pinecone isn't just fast—it’s essential when working with large-scale generation
With the right AI stack, we can turn imagination into shoppable reality