Your question is Production Metrics for AI Agents. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
What are the key metrics you would define to evaluate the performance of an AI agent in production?
Explain how you would organize the metrics across task success, response quality, operational performance, safety, and business outcomes. Distinguish metrics that can be computed automatically from those requiring human review, and describe how you would establish baselines, monitor regressions, and handle non-deterministic behavior.
