What is RAG (Retrieval-Augmented Generation)?
RAG is a technique where a model retrieves relevant documents at query time and generates its answer from them instead of from memory alone.
Published
Definition
RAG (Retrieval-Augmented Generation) is a technique in which a system retrieves relevant documents at the moment a question is asked and generates the answer from those documents, rather than from the model’s training alone.
Almost every AI assistant that cites sources is doing some form of RAG.
Why It Matters for AI Visibility
RAG is the reason AI visibility is winnable at all. If models answered only from training, your content would need to have been in a dataset frozen months ago and there would be nothing to optimize. Because retrieval happens live, a page published today can be cited today.
It also explains what to optimize for. The pipeline has stages, and your page has to survive each one: it must be crawlable, then chunked into passages that stand alone, then judged similar enough to the query, then trusted enough to be quoted. Failing any single stage produces the same outcome as failing all of them.
Example
A typical pipeline:
- User asks a question
- The system embeds the question as a vector
- It searches an index for the closest passages
- The top passages are inserted into the model’s context
- The model writes an answer and cites the passages it used
Your content enters at step 3 — as a passage, not as a page.