wtf is simple-rag-agent?

Simple RAG Agent is a tutorial-style project that teaches you how to build an AI assistant that can both answer questions from your own documents and perform calculations on demand. It runs as a Jupyter Notebook, meaning you can read through it and run the code cell-by-cell. The practical benefit: instead of a chatbot that only knows what it learned in training, you get one that reasons through problems step by step using your data and tools you give it. The project combines two ideas. First, Retrieval-Augmented Generation (RAG), which lets an AI model read your documents and answer questions about them. Your files get loaded, broken into chunks, and converted into numerical representations called embeddings so the system can find relevant passages by meaning rather than keyword matching. Second, it adds an "agent" layer, a ReAct agent that can think through a problem, decide which tool to use, take action, and explain its reasoning. The demo includes two simple Python tools: one calculates factorials and the other checks if a number is prime. The agent decides when to call each one. Someone learning how to build AI applications would use this as a hands-on starting point. For example, a founder building a medical-info chatbot could study how the notebook's lung cancer questions pull relevant context from documents while also computing answers (like the factorial of treatment cycles). A product manager exploring what AI agents can do could run the notebook to see an agent reason through a multi-step question live. The project uses LlamaIndex for the RAG pipeline and Ollama to run an open-source language model locally, so you don't need a paid API key. The tradeoff is that running locally requires decent hardware and is slower than calling a hosted model. The README doesn't go into detail on setup prerequisites or how to get Ollama running, so some outside familiarity is needed.