Retrieval-augmented generation (RAG)
An AI architecture that retrieves relevant documents from an index or the web and feeds them to a language model before it writes an answer.
RAG pairs a search step with a writing step. The system turns a question into a query, retrieves the most relevant passages, and gives them to the model as context. The model writes an answer from that material and can point back to it. The name comes from a 2020 research paper, and the pattern now sits behind AI search experiences generally.
The RAG pipeline in plain terms
- Rewrite. The question becomes one or more search queries.
- Retrieve. A search index returns candidate pages or passages.
- Rank. The system keeps the passages that look most useful.
- Generate. The model writes an answer from those passages.
- Cite. The answer links the sources it drew on.
Why AEO is mostly a retrieval problem
Every step is a chance to drop out. If you're not in the index, you're never retrieved. If your page buries the answer, a better-structured passage wins the ranking step. If your passage is vague, the model paraphrases someone else. That's why grounding, crawl access and answer-first content matter more than tricks aimed at the model itself.
Somewhere, right now, an AI is recommending your client's competitor.
Elite Answers is invite-only. Request access through Elite Digital and the team will set you up.