Retrieval-Augmented Generation, usually shortened to RAG, is a technique used to make AI language models more accurate and up to date.
The Core Idea
A standard AI language model generates answers based only on what it learned during training, which has a fixed knowledge cutoff. RAG adds a retrieval step: before answering, the system searches an external source, such as a document database or the web, and feeds relevant snippets into the model alongside the question.
Why It Helps
Because the model now has fresh, specific information in front of it, RAG can reduce factual errors and allow chatbots to answer questions about recent events or private company documents that were never part of their original training data.
Where It Is Used
RAG is common in customer support chatbots, internal company knowledge assistants, and AI-powered search tools, anywhere accuracy and up-to-date information matter more than pure creativity.