AI-Powered Visual Shopping Assistant

**AI-Powered Visual Shopping Assistant** – a web-based AI tool that helps users find and buy products by simply uploading a photo.
Point it at a room, a person, a pet, or any object — and it returns a curated list of matching products with direct purchase links.

**Problem**
People constantly see styles they love — in a room, on a person, or in a photo — but have no way to find the actual products behind it. Search engines don't understand visual context: typing "blue vase" returns thousands of irrelevant results. Hours are wasted browsing and comparing with no real guidance. Decision fatigue kills the purchase before it even happens.

**Solution**
A simple web platform where users upload any photo. The AI vision model analyzes the image — detecting style, colors, and context — and returns a curated list of matched products with direct purchase links. No typing. No searching. One photo — instant, actionable results.

**Simplified Cases**
1. A user uploads a photo of their living room. SnapShop suggests furniture, rugs, and decor items that match the existing color palette and style — all with buy links.
2. A user uploads a photo of their dog. The tool recommends collars, beds, outfits and toys matched to the animal's size, breed, and color.
3. A user uploads a photo of a Christmas tree. SnapShop finds the exact ornaments, lights, and trimmings to complete it.

**Tech Stack**
Frontend — React (single-page web app, photo upload UI)
Backend — FastAPI (orchestrates the flow, calls AI, returns JSON results)
AI Vision API — GPT-4o / Gemini (analyzes image, generates product queries)
Shopping Data — Google/Amazon API (fetches real products with prices and buy links)

The issue names no repository files, tests, or entry points, so start by inspecting the repository structure and existing application setup. The scope spans photo upload, AI vision, product search, and purchase links; a concrete implementation boundary and acceptance criteria are still needed to define when it is done.

Written by the indexing model from the issue text.