Core Concepts
Semantic Search in plain English.
Also known as: meaning-based search,neural search,vector search
The one-sentence version
Search that matches on meaning rather than exact words, usually by comparing embeddings of the query and the documents.
Semantic search returns results that are about the same thing as the query, not just results that contain the same words. It works by converting the query and every document into embeddings, numeric representations of meaning, and finding the documents whose vectors sit closest to the query's. Ask "how do I stop being charged" and a semantic search will surface a page titled "Cancelling your subscription" even though no words overlap. It powers the retrieval step in RAG, the "related items" features in many apps, and the AI search tools that have replaced keyword search in products like Notion and Slack. Its weaknesses are the mirror image of its strengths: it can miss exact identifiers, it depends on the embedding model understanding your domain, and it returns something even when nothing relevant exists. That is why production systems usually pair it with keyword search (hybrid search) and a reranker, and why the quality of the embedding model matters as much as the database that stores the vectors.