Semantic Search with Vector Embeddings
How It Works
- Your search query is converted into a semantic representation (embedding) using AI
- This embedding captures the semantic meaning of your query
- The system compares it with customer data embeddings in the database
- Results are ranked by similarity score (higher is better)
Technical Details:
- Computes Cohere multi-lingual embeddings via Heroku Inference for vector similarity search
- Leverages Heroku pgvector for vector storage and similarity search
- Uses the popular Northwind database for an example database. View this Dataclip to see all customer information when crafting your search queries