In plain language
It is a search index for meaning: give it a question’s embedding and it returns the closest pieces of content.
Example
Help articles are split into chunks, embedded, and stored in a vector database that an AI assistant searches before answering.
Related terms
Embedding · Semantic Search · Retrieval-Augmented Generation (RAG) · AI Memory
Common misconception
A dedicated vector database is not always necessary; many existing databases, such as PostgreSQL with pgvector, can store and search embeddings.
Implementation guidance
Store source links and metadata with each vector, use the same embedding model for content and queries, re-embed when content changes, and monitor result quality.