Pipeline
Every stage the audio passes through, with time, model and cost. Click a stage for its raw event.
Run a case to populate the pipeline.
Speaker attribution (diarization)
Transcription accuracy (WER)
Evaluation — how good is the judge?
For us and procurement, not the reviewer: the AI’s findings scored against
the errors we deliberately seeded into this hearing.
Language capability tiers
Which languages the AI will support, assist-only, or decline outright.