Vector Database

Working knowledgeAgentic Protocols and Infrastructure

Also called: Vector Store

A vector database is a database built to store embeddings and find the nearest matches to a query fast enough to sit inside a live application. A conventional database finds records matching a value exactly; a vector database finds the records closest in meaning to the question asked. It is the retrieval layer of a RAG system: documents are embedded once and held here, then each incoming question is embedded and used to pull the nearest passages back.

In practice

A vector database is a second copy of your content, which makes it a data governance object as much as a technical component: it lives somewhere, under someone's control, and falls under the same residency and access rules as the source. Ask where it is hosted, who can query it, and how a deletion in the source system reaches it.

Not sure where your organisation stands?

Take the free AI-readiness diagnostic.

Start the diagnostic