Semantic search
Search that matches by meaning, not keywords — type "how to cancel" and find the doc titled "ending your subscription."
Think of it like
A librarian who understands what you're looking for, not just the exact words you used.
Example
Search your docs for "pricing" and get results about "cost," "rates," and "plans" — because they mean similar things.
How it actually works
Your query and every document chunk are turned into embedding vectors. The search engine finds documents whose vectors are closest to the query vector (by cosine similarity). It's why AI-powered search "just gets it" compared to keyword search.
For product teams
Search that understands intent, not just keywords — the UX leap users expect now.
For engineers
Query and documents embedded in shared vector space; ranked by cosine similarity.
Related
- Embeddings — Powered by embeddings under the hood.
- Stored and queried in a vector database.
- RAG — The retrieval mechanism 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