Logs AI agent decisions with reasoning, confidence scores, and alternatives for debugging.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.