#grounding

Retrieval-Augmented Generation (RAG)

Grounding LLM responses in chunks retrieved from an external corpus so the model reasons over real, citable sources instead of parametric memory alone.