Overview
Vector databases are optimized for 'similarity search' rather than exact matches. They allow systems to find pieces of information that are semantically related to a query.
Use in RAG
In a RAG system, documents are converted into vectors (embeddings) and stored in a vector database. When a user asks a question, the system finds the most similar document vectors to provide context to the LLM.
Popular Examples
- Pinecone
- Milvus
- Weaviate