Vector database

A vector database stores embeddings — the numeric meaning of text — and finds the closest matches quickly. It is the storage behind meaning-based search and many "chat with your docs" tools.

You rarely touch one directly, but it is what lets an AI assistant pull the most relevant passages from thousands of company documents before answering.

Thinking it stores answers. It stores material to retrieve; the model still has to read and use that material correctly.

From our knowledge base, retrieve the three most relevant sections to this question and answer using only those.