What it means
A vector database is a database designed to store embeddings and quickly find the ones most similar to a given query. It is a common component of semantic search and RAG systems, usually alongside the original text and its metadata.
Real-world example
A health authority converts 15,000 pages of clinical guidelines into chunks and stores an embedding for each in a vector database, together with metadata such as guideline title, version and publication date. When a staff member asks a question, the system finds the most similar chunks in milliseconds and passes them to the language model.
Why it matters for AI adoption
A vector database is infrastructure that needs to be secured, backed up and kept in sync with source documents. It also stores (or links to) the organisation's content, so access controls and data protection rules apply. Some existing database and search systems now offer vector search, so a separate product is not always necessary.
Often confused with
The embeddings are the data; the vector database is where they are stored and searched.
Questions to ask
- Do we need a separate vector database, or can our existing systems handle vector search?
- How are deleted or updated source documents removed or updated in the vector database?