Similarity Search
Finding the items whose vectors are closest to a query’s vector — the engine under semantic search.
Think of it like
Finding the nearest coffee shops to your location — sort everything by distance, take the closest few.
Example
The query becomes a vector, and the system returns the ten stored vectors sitting nearest to it in the space.
How it actually works
Similarity search is the mechanical core: given a query vector, find the nearest stored vectors by a distance measure like cosine or dot product. Done naively it means comparing against every vector, which is too slow at scale, so real systems use indexes that trade a little accuracy for huge speed. It’s the same operation whether you call the result semantic search, recommendations, or memory retrieval.
For product teams
It’s the shared engine behind semantic search, recommendations, and agent memory — one primitive, many features.
For engineers
Return the k nearest vectors to a query under a distance metric, usually via an approximate index for speed.
Related
- Semantic Search — The user-facing feature it powers.
- Vector Index — The structure that makes it fast.
- Embeddings — What gets compared.
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