agentic-rag-for-dummies

Why we included this project

Most RAG tutorials stop at retrieving a passage and pasting it into a prompt, so this repository is a useful next step for anyone who wants a retrieval system that actually reasons as it searches. Built on LangGraph, it walks through an agent with conversation memory that rewrites ambiguous queries, asks for clarification when needed, fans out parallel retrievals for multi-part questions, and retries when the first results miss. Each layer, from the LLM provider to the embedding model and indexing strategy, can be swapped independently, which makes the project a practical starting template rather than a fixed demo. It defaults to local Ollama models, with examples for OpenAI, Anthropic, and Google, so you can try it on your own hardware before spending anything on API calls. Beginners get the most from the step-by-step notebook, while teams that already know the basics can lift the modular structure straight into a prototype.