Review a conversation outcome

After a test or customer call, open the assistant's conversation history and select the matching record. Use its timestamp and identifier to distinguish repeat attempts.

Check completion and content separately

A completed run means the execution finished. To decide whether the request was resolved, compare the transcript, collected values, tool responses and any configured quality checks. Recording and transcript availability can depend on the storage setup and post-call processing.

Compare test and production calls

Browser tests exercise the voice setup without a phone number. A carrier call also depends on routing, account credentials and audio transport. Test both when the service will be used on a telephone line. Draft tests can use different instructions from the published version handling customers.

Follow the result into another system

Check delivery at the webhook receiver or retrieve the record with API v2. Use the conversation ID when correlating events; names and timestamps alone may not be unique.