Detects LLM hallucinations with source grounding, confidence scoring, and fact verification.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.