RAG is a method that takes a user's question, searches for relevant information, and then combines the retrieved info with the question to prompt an AI chat model. It has two main stages: preparing and indexing information, and dealing with questions. Postgres plays a key role in RAG, often used as a vector database, and pgrag is an extension that helps create a complete RAG pipeline without leaving psql. pgrag provides features such as text extraction, chunking, and embedding generation for both stages of the pipeline. The pipeline uses a vector database to store document chunks and their corresponding embeddings, allowing for efficient searching and retrieval of relevant information. With pgrag, users can create a complete RAG pipeline quickly and easily, using SQL queries instead of writing custom code. The extension is available on Neon right out of the box or can be compiled for use with other Postgres servers.