Every call is recorded and transcribed, then scored by an LLM judge against your quality rubric. Open a call to play the recording, read the transcript and see the event timeline.

Review queue

Calls scored below your org threshold (or flagged by an operator) land in the Review queue. Reviewers approve, send to training, or dismiss — the 'needs training' bucket feeds the fine-tune builder.

Quality rubric

The rubric defines the axes, weights and per-axis instructions the judge uses. Changes apply to future calls (Settings → Quality rubric).