Builds AI evaluation pipelines with metrics, human review sampling, baselines, and statistical checks.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.