RAG is the architecture where a model retrieves relevant documents first and generates its answer from them — the mechanism behind grounded, citation-backed AI answers.
Instead of answering from parameters alone, a RAG system searches (the web, a corpus, a database), stuffs the best passages into the model's context, and asks it to answer from those. The output inherits the retrieval: change what is found, and you change what is said.
RAG is why GEO is tractable at all. You cannot edit a model's weights, but you can absolutely influence the documents a retrieval step finds — that is a content and PR problem, and brands know how to work those.