What it means
Retrieval-augmented generation, or RAG, is a simple idea with a clumsy name. Before the assistant answers, it retrieves current sources, searching the web or a document store, and then writes its reply from what it found rather than from memory alone. The answer is grounded in pages it just read.
Why it matters to you
A model's built-in knowledge is frozen at training time and can be months old. RAG is the opposite: it reads the live web. That is why a clear, crawlable, quotable page can change what an engine says about your business today, with no retraining in between. The pages an engine retrieves are the pages it can quote and cite, so being retrievable is the first step to being named.
Which engines retrieve
Perplexity and Google AI Overviews retrieve by default, which is why they show their sources. ChatGPT, Claude and Gemini retrieve when they browse for a current or local answer instead of replying from memory.
How to be retrieved
Make your pages easy to fetch and easy to quote: let AI crawlers reach them, answer the question near the top, and back it with structured data for AI. A page that is easy to retrieve is easy to turn into an LLM citation. It is the mechanism underneath answer engine optimization, and our guide on how home services get recommended by AI shows it in practice.
See whether AI retrieves you
A free AI Visibility Snapshot shows how AI names you plus three competitors, across ChatGPT, Claude and Gemini.
