Vector databases are specialized in storing high-dimensional vectors efficiently and finding the nearest neighbors to a query. This allows semantically similar documents to be identified even when they share no common keywords. Well-known solutions include Pinecone, Weaviate, Qdrant, and pgvector as an extension for Postgres. In RAG systems they supply the relevant passages that are given to the model as context.
Vector Database
A vector database stores embeddings and enables fast similarity searches by meaning rather than keywords. It is the foundation for semantic search and RAG applications.