Vector Database

A vector database stores AI embeddings — numerical representations of text, images, or other data — and lets you search them by similarity rather than by exact match. It's the backbone of every RAG system, semantic search engine, and AI-powered recommendation. Major vector databases in 2026: Pinecone (managed SaaS), Weaviate (OSS + managed), Qdrant (OSS, Rust-based), Chroma (lightweight), Milvus (enterprise), and pgvector (Postgres extension — increasingly the default for teams already on Postgres). The category consolidated heavily in 2025-2026; the winners are now those that combine vector + keyword + filter search (hybrid search) and run at production scale. Distinct from but often confused with traditional databases — vector DBs don't replace them, they sit alongside.