What Is RAG?

RAG stands for Retrieval-Augmented Generation and allows AI to find and use relevant information when answering a question, rather than relying only on what it already knows.

Imagine asking an AI about your company’s holiday policy; with RAG it could search your HR documents, find the relevant information and use that to answer your question.

Why Is RAG Useful?

Businesses have huge amounts of information that an AI model won’t automatically know about, from internal policies and product information to support guides and company documents. RAG gives the AI a way to find what’s relevant to the question being asked and use it as part of its response.

This can make the answers much more useful and specific to an organisation, while also reducing the chances of the AI making something up. It doesn’t guarantee that every answer will be correct, because the AI still needs to find and use the right information, but it gives it something more relevant to work with.

A useful way to think about RAG is that instead of expecting the AI to know everything, you’re giving it somewhere to look.