Retrieval-Augmented Generation (RAG)
AI Technical
An architecture that grounds an LLM’s responses in retrieved documents at inference time. Reduces hallucination and keeps sensitive data outside training, but introduces privacy obligations on the retrieval corpus itself.