How It Works
Learn how the Product Mapping Tool uses AI to find similar products and analyze images
Upload Your Data
Start by uploading CSV files containing your product data. The system supports various product information formats.
- CSV file upload with validation
- Automatic data cleaning and formatting
- Support for multiple product categories
- Real-time upload progress tracking
AI Image Analysis
Use OpenAI Vision to analyze product images and extract detailed product information automatically.
- Customizable analysis prompts
- Brand and product type identification
- Confidence scoring for accuracy
- JSON format output for easy processing
Vector Embeddings
Convert product data into high-dimensional vectors using OpenAI's embedding models for semantic similarity matching.
- Text embeddings for product descriptions
- Image embeddings for visual similarity
- Combined embeddings for comprehensive matching
- Pinecone vector database storage
Find Matches
Search for similar products using semantic similarity. The AI finds products that are conceptually related, not just textually similar.
- Semantic similarity search
- Configurable similarity thresholds
- Multiple search modes (text, image, combined)
- Real-time results with confidence scores
Complete Workflow Process
Data Upload
Upload CSV files with product data through the Upload Data page
Data Processing
System automatically cleans and validates the uploaded data
Embedding Generation
AI creates vector embeddings for text and image data
Vector Storage
Embeddings are stored in Pinecone vector database for fast retrieval
Similarity Search
Use Find Matches to search for similar products using semantic similarity
Results & Analysis
View detailed results with confidence scores and similarity metrics
Technology Stack
FastAPI
Modern Python web framework for building APIs
OpenAI API
GPT-4 Vision for image analysis and text embeddings
Pinecone
Vector database for storing and querying embeddings
Supabase
PostgreSQL database for storing product metadata
Jinja2
Templating engine for dynamic HTML generation
Uvicorn
ASGI server for running the FastAPI application