Fetching real documents and putting them in the prompt.
Retrieval-augmented generation searches a document store for passages relevant to the question and pastes them into the model's context before it answers.
It is the standard fix for stale or missing knowledge: the model does not need to have memorised your policies, it just needs to read them at request time.
It also makes citation possible — the answer can point at the retrieved passage — which is why it underpins most enterprise deployments.