also vector, embedding model
A list of numbers that places a piece of text so that similar meanings sit close together. Good for finding related text; it says nothing about whether that text is correct, current or allowed.
You have done this if
Semantic search found the right policy for a paraphrased question that keyword search missed, and in the same week confidently returned an outdated one.
Say it in a review
Embeddings give us candidates by meaning; freshness and permissions are filtered separately.
On the AI Application map Vector Index, Retrieval
Read Similar is not the same as correct: the limits of embeddings · A vector store is infrastructure, not magic memory