3.3.1. Vector Embeddings for RAG

Embeddings convert text into numerical vectors (arrays of floats) that capture semantic meaning. The key insight: similar concepts produce similar vectors, even if they use different words. "Car" and "automobile" have nearly identical embeddings despite being different strings.

Why embeddings matter for RAG:

• Enable semantic search (find conceptually similar content, not just keyword matches)
• Power vector databases for efficient similarity lookups
• text-embedding-ada-002 produces 1536-dimensional vectors

Embedding workflow:

1. Chunk your documents into manageable pieces (~500 tokens)
2. Generate embeddings for each chunk
3. Store chunks + embeddings in a vector database
4. At query time: embed the query, find nearest chunks, include in prompt