loading…
loading…
Motion AI logged 17 action items and named an owner for each. BP listened for whose voice carried the commitment. Of the 15 that were actually spoken aloud, 7 were carried by a different voice than Motion AI named — and 2 of Motion AI's 17 were never voiced on the call at all. Click any ▶ to hear the moment.
BP conviction scores and acoustic flags populate once N=4 speaker diarization completes. Motion AI captures and estimated timestamps shown below now.
| # | Action item | 🤖 Motion AI named | 🎙️ BP heard (real voice) | ▶ Evidence |
|---|
ISIA on one side, OSA on the other. Each speaker gets a six-dimension acoustic profile, a conviction score, and a quote bank tied to audio playback.
Beyond per-speaker analysis: how the 4 voices interacted, where conviction concentrated, where alignment broke.
Speaking time as a fraction of total meeting.
Average emotion scores aggregated by side. The OSA side projects maximum-confidence selling mode · the ISIA side carries roughly 2x the concern signal. Healthy pitcher-decider asymmetry · flat parity would be a red flag.
The marquee Beyond Physician findings · moments the meeting recap cannot surface.
Per-speaker fillers per minute · plus top three filler patterns. David's "like" concentrates in the 20:00 equity explainer; Patel's "right" is confirmation-seeking, not hesitation.
First utterance per topic · indicator of who steered the meeting from section to section. Patel opened 6 of 13 · he wasn't just dominant in speaking time, he set the topic flow.
David ran the SELL · Patel ran his own PRACTICE OPS · two parallel meetings in one.
Verbatim transcription + proprietary acoustic conviction analysis + dimension-mapped quote extraction. Zero third-party APIs. All MIT/ISC/BSD components.
Every score is a weighted blend of three independent layers. The split is the patent-pending product of years of voice research on healthcare interview data.
| Phase | Step | What happens |
|---|
| Dimension | Acoustic signature |
|---|