Retrieval-Augmented Generation (RAG)

A language model's training data has a cut-off date and cannot include private or fast-changing information by default. RAG addresses this by adding a retrieval step: when a question comes in, the system searches a knowledge source — a website index, an internal document library, a live database — for the most relevant passages, and feeds those passages to the model alongside the question so its answer is grounded in current, specific material rather than a general trained impression. This is the mechanism behind AI browsing and search features in tools such as Perplexity, Copilot and Google's AI Overviews, and behind most enterprise AI assistants built on a company's own documents.

RAG is directly relevant to how businesses get cited by AI, because the documents an engine retrieves and quotes are exactly the pages that need to be crawlable, well-structured and factually clear — which is the practical overlap between RAG and generative engine optimisation. It is also the architecture most AI consultancy engagements use when building an assistant that needs to answer accurately from an organisation's own policies, catalogues or knowledge base rather than hallucinating.