Analyzes AI outputs and suggests ways to reduce hallucinations and improve reliability.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.