Vector Database

A database optimized for storing and searching high-dimensional numerical vectors. In agentic systems it's used to find semantically similar documents quickly — the backbone of most RAG implementations.

📌 Example: You store embeddings of your docs in Pinecone or pgvector. At query time, the agent's question is converted to an embedding and the database returns the closest matches.

🛠 In practice: Similarity is measured by distance (e.g. cosine similarity), not keyword matching, so results are conceptually relevant even if they don't share exact words with the query.

• depends on RAG (Retrieval-Augmented Generation)RAG patterns typically rely on a vector database to find semantically relevant documents at query time.
• depends on EmbeddingA vector database stores and searches embeddings — you need embeddings to populate and query one.