Retrieval augmented generation (RAG) is a strategy that helps address both LLM hallucinations and out-of-date training data.