Ask for an explanation backed by the execution trace
Explain this run turns retained trace evidence into observations, possible explanations, and next steps, with citations you can inspect.
Click Explain this run from an execution timeline to get an AI-generated explanation grounded in the trace. It separates observed facts from possible explanations and suggested next steps. Each citation opens the supporting event so you can check the reasoning.
The explanation discloses missing or truncated evidence and respects the run's logging and redaction policy. It does not change the run, execute a proposed fix, or decide whether the agent met your business requirements.
Analysis runs only when you request it and uses your account's model access and normal usage charges. You can still choose to capture an agent run as an eval case yourself.