Grounding means tying a model's answers to authoritative source material rather than its own memory.
By giving Claude the relevant documents, often via retrieval or MCP resources, and asking it to answer from them, you reduce hallucination and improve accuracy. Grounded systems can cite where an answer came from. It is a key technique in retrieval-augmented generation and agents.