Vector Database
Basis Data Vektor
Storage and search built for embeddings rather than exact values.
A vector database stores embeddings and answers nearest-neighbour queries: given a query vector, return the stored vectors closest to it. Under the hood it usually uses an approximate nearest neighbour index, trading a little recall for speed on millions of vectors.
Most products today do not need a dedicated vector database. Postgres with a vector extension, or plain full-text search with a reranking step, covers a lot of ground and keeps the data in one place. A dedicated engine earns its cost when the corpus is large, the search is central to the product, and retrieval quality matters more than operational simplicity.
Related terms
Embedding
A list of numbers that places text near other text with similar meaning.
RAG
Fetch the relevant documents first, then let the model answer using them.
Semantic Search
Search that matches on meaning rather than on the exact words typed.
Retrieval
The step that finds the material a model will answer from.
Want this working inside your team?
It starts with one conversation about your workflow.