Eight reads from one person on video.

Composure Index reads decision readiness under pressure, including stress recovery speed. Presence Quotient reads nonverbal authority from camera engagement, posture, and smile markers. Cognitive Clarity tracks mental bandwidth from blink variance. Decision Clarity Window marks peak clarity and sustained focus. Congruence Signal reads how consistently the signals on a face agree with each other, and returns no verdict on whether a person is genuine. Culture Erosion Index surfaces team friction as a discussion starter. Resilience Arc tracks recovery. Momentum Signature reads whether focus is achieved, held, and self reinforcing. When the footage cannot support one of them, the engine abstains and reports it as not measured.

Every score traces back to measurable facial geometry, on published methods. Larger files also add voice and language on the server path. Each report shows three layers: the raw measurements the model detected, how they changed over time, and the interpretation drawn from them. So if a finding reads as strange, you trace it back from insight to pattern to raw number.

No. The scores are decision support, not a verdict on any person. They are not built for hiring, firing, or promotion calls. The Culture Erosion Index, for example, is framed as a discussion starter, not an assessment. Read the numbers as evidence alongside your own judgment, never as a substitute for it.

Not the footage. On The Read the video stays in your browser. The server path runs when you turn on room mode, add a second camera, or your browser cannot decode the file. There the footage is deleted as soon as the analysis finishes, and within 60 minutes if the run is abandoned or fails. We keep the derived numbers, not the video, and those numbers stay in your report.