What is RAG?

Definition: RAG (Retrieval-Augmented Generation) is a technique that gives a language model access to external documents at answer time. Instead of relying only on what it memorised during training, the model retrieves relevant information first, then generates an answer using it.

The one-sentence idea

RAG = "look it up, then answer." Rather than hoping the model already knows something, you fetch the right text and hand it to the model along with the question.

Why it matters

• Answers are grounded in real, current, specific information
• The model can use your private data (company docs, a product manual) it was never trained on
• It dramatically reduces made-up answers (hallucinations)

RAG is the technology behind "chat with your documents", customer-support bots that quote your help centre, and most enterprise AI assistants.

💡 Prerequisite: a little Python helps for the runnable demos, but the concepts work for anyone. Some lessons include live code you can run.