RAG combines a language model with a live retrieval step. Instead of answering only from training memory, the system searches a source — the web, a database, your site — pulls the most relevant passages, and writes its answer from them. Most AI search tools work this way.
This is why fresh, well-structured content matters: RAG can only use what it can retrieve and parse. If your page is the clearest match for a query, it becomes the source the model builds its answer on.
Ask Perplexity a question and it retrieves a handful of pages, reads them, and composes an answer citing each — classic RAG in action.
Why this matters for AI findability
RAG is the mechanism that lets AI quote your current content rather than stale training data. Being retrievable and quotable is precisely what agentic findability optimises for.