Vector database
A database built to store meanings as coordinates and find the nearest ones fast — the memory bank behind RAG.
Think of it like
A library shelved by what books are about, not by title — so “nearby” means “similar in meaning.”
Example
Ask a support bot a question and it pulls the three closest help-doc chunks from a vector database.
How it actually works
It indexes high-dimensional vectors and answers “what’s closest to this?” in milliseconds using approximate nearest-neighbor search. Exact search is too slow at scale, so it trades a sliver of accuracy for big speed. The index structure (like HNSW) is what makes that trade work.
For product teams
The storage layer that makes semantic search and RAG feasible at scale.
For engineers
An ANN index over embedding vectors supporting top-k similarity queries with metadata filters.
Related
- Embeddings — Stores the embeddings that power retrieval.
- RAG — The lookup step inside RAG.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome