What is RAG?

A technique for giving AI access to external documents before it responds.

// Explanation

RAG (Retrieval Augmented Generation) is the solution to one of the biggest problems with LLMs: they only know what they learned during training. With RAG, before responding, the model searches a custom document base (your manual, your knowledge base, your policies) and uses that information to give precise, up-to-date answers.

// Example

A support chatbot that answers questions from your product manual using RAG. Without RAG, it would make up answers. With RAG, it cites the manual.

← See all 42 glossary terms