Retrieval-augmented generation is a method in which a language model first fetches relevant documents and then writes its answer using them as context.
The approach was named in a 2020 paper by Patrick Lewis and colleagues at Facebook AI Research. Instead of relying only on what the model memorised during training, the system searches an index at question time and hands the best passages to the model.
Most AI search products work this way. That is why fresh, well structured pages can show up in answers soon after they are published, long before any model is retrained.
For publishers the lesson is practical: you are writing for a retriever first. Clear headings, self-contained passages and accurate titles decide whether your page is among the few the model ever sees.