How to build a local retrieval-augmented generation application using Postgres, the pgvector extension, Ollama, and the Llama 3 large language model.