Glossary
Retrieval-Augmented Generation (RAG)
Retrieval-augmented generation (RAG) is a technique where a language model retrieves relevant documents from an external source at query time and uses them to ground its answer. It pairs a search step with the generation step so responses can cite current, specific sources.
RAG is the mechanism behind most cited AI answers. Rather than relying only on what a model learned in training, the system searches an index, pulls matching passages, and conditions the answer on them. That retrieval step is where a brand's pages can enter an answer as a source.
For visibility work, RAG explains why page structure matters so much. If a page is easy to retrieve and holds a clean, self-contained passage answering the query, it is more likely to be pulled into the context the model generates from, and then cited.
Common questions
Why does RAG matter for AI visibility?
Because the retrieval step decides which pages reach the model when it answers. A page that is easy to retrieve and holds a direct, self-contained answer is more likely to be used and cited, which is what visibility work targets.
Is RAG the same as a model's training data?
No. Training data is what the model learned in advance. RAG fetches documents live at query time and grounds the answer on them, which is how answers can reference current and specific sources.