What is a vector database and why is it used in AI applications?

A vector database stores data as high-dimensional vectors (embeddings) and enables fast similarity search.

**Why it's needed:**
LLMs understand meaning, not exact keywords. To find 'similar' content, you need to compare meaning — which requires vector similarity search.

**How it works:**
1. Text → Embedding model (e.g., OpenAI text-embedding-3-small) → vector [0.12, -0.34, 0.87, ...]
2. Vectors stored in vector DB
3. Query → same embedding model → query vector
4. DB finds most similar vectors (cosine similarity)
5. Return matching documents

**Popular vector databases:**
- **FAISS** (Facebook) — local, fast, free
- **Pinecone** — managed cloud, easiest to use
- **Weaviate** — open source, self-hosted
- **Chroma** — best for LangChain integration
- **pgvector** — PostgreSQL extension

**Use cases:** RAG chatbots, semantic search, recommendation systems, duplicate detection