Evaluators

support-agent

Automatic checks that score new runs as they arrive.

Evaluators
NameChecksRuns onStatus
Correct answerAnswer matches the tool results10% of productionRunning
TonePolite and clear10% of productionRunning
Refund policyRefund follows policyAll refund-agent runsRunning
Hallucinated order idOrder ids exist in tool outputAll runsPaused

Sample content. Create a free project to build your own. →