Find the reported failure
Search captured phrases and filter by source, model, tool, or status.
Find out why an agent failed.
Agent Observability. Trace a response through the captured model and tool calls behind it. Investigate errors and give your team a concrete case to evaluate.
Agent Observability. The calls behind the response.
Inspect the calls
Review available inputs, outputs, errors, timing, and usage. Keep missing measurements visible.
Monitor recurring problems
Review quality monitors and alerts, then open an investigation or prepare an evaluation case.
Include external agents
Send traces from external applications using OpenInference. Keep their reported observations identifiable.
One failure. A better next version.
Follow a support agent’s unverified refund promise from captured evidence to a reviewed case and a comparison of two releases.
Locate the captured refund promise and the calls around it.
Read what the application reported and distinguish the response from a confirmed refund.
Keep the original request and supply a reviewed expectation: check the policy and request approval.
Inspect the same case on the baseline and candidate, including their different responses.
One candidate result is missing. The comparison remains inconclusive and calls for more evidence.
In practice
A support agent says a refund has been issued. The captured evidence shows no confirmed refund. The team inspects the policy call, records the failure, and prepares an evaluation case that requires approval before a refund promise.
Explore the solutionDiscuss an agent failure.
Bring an agent failure your team needs to understand. We’ll discuss the calls, logs, and business records needed to investigate it.