Local RAG Agent

The purpose of this repo is to demonstrate how to build a basic local RAG agent. This could e.g. be used for making internal company documents "chattable".

The full demo can be seen here.

Brief tech stack overview

• The Ollama official docker image is used managing and running the language models.
• We deploy Meta's Llama3.2 large language model and Nomic's nomic-embed-text embedding model.
• Weaviate is used as vector store.
• LangChain is used for orchestration of the RAG agent.
• The agent is deployed locally with the LangGraph server.
• Finally, the Agent Chat UI app is chosen as web UI.