Retrieval-augmented generation
Also called: RAG
In one sentence
Fetching relevant documents and putting them in the context so the model answers from real sources.
In more depth
Rather than relying on what a model absorbed during training, the system searches a collection of documents, selects the relevant passages and includes them in the prompt.
This is the standard way to make a model answer about private or current material, and it reduces — but does not eliminate — fabrication, because the model can still misread or over-extend what it retrieved.