Operate and Improve

Investigate a Platform run, use supported operator controls, reconcile incomplete effects, and turn observed failures into repeatable evaluations.

Start with the workspace, run or conversation identifier, approximate time, and the outcome being investigated. Use a credential permitted to inspect that workspace. Operator actions require additional permissions and channel support.

Use the channel’s lifecycle controls, then turn a reproduced failure into a candidate evaluation.

Find the Execution

In the standard Console workspace, open Evaluate → Runs. Operator-only workspaces retain Operate → Runs. The unified inventory includes supported conversation and framework run records; the old Conversations listing redirects there. Conversation detail remains the place to inspect the underlying conversation. The Runs API documents filtering, detail, and supported controls.

Record the selected service/version, channel, execution state, and available error or correlation identifiers. Read history and traces as evidence with their stated completeness limits. A generated summary or missing recording link is not sufficient to determine whether an action completed.

Choose the Correct Lifecycle Action

Observation	Next action and verification
An application text conversation has reached its intended outcome	Apply the documented conversation close behavior and read back its state; preserve the durable answer before cleanup
A persistent text connection drops	Follow the session protocol; verify the conversation state before creating another interaction
A managed message thread is complete	Use that channel's lifecycle; thread identity, terminal closure, and a forced new conversation are separate decisions
An eligible voice run needs human intervention	Use your operator API integration, verify takeover, and verify handback separately; voice operators join by phone
An integration action is approved or a request times out	Inspect the target system before retrying; approval alone does not confirm that the action completed

Do not apply one channel's close or retry behavior to every session type. Where an outcome is ambiguous, preserve identifiers and reconcile before repeating a mutation.

Recover and Escalate

Follow Troubleshoot an Integration to locate access, configuration, execution, projection, and external-delivery failures. Record the last successful step, the observed error, and the next check. Share only the information needed to investigate through the agreed support channel.

Improve

Review Home → Insights for patterns across runs. Find Quality and Dashboards through Console search; these pages remain available according to workspace access even though they are not standard sidebar entries. Use quality metrics, production evaluations, and trace analysis according to their access and coverage limits.

Turn a reproduced failure into a test case with a stated expected result. Test the proposed configuration against that case using Build and Prove. Record cases that could not be evaluated separately from passes, and check model-generated assessments against the underlying run and target-system records.