One engine.
Every room.
A game. A boardroom. A keynote. A call. The engine reads the behavioural signals people give off but cannot name, and returns objective scores.
Full body pose tracking. Pressure response, read from game film.
Illustrative engine panel. The footage is real. The scores are illustrative examples, not a live measurement. It cycles sport, business, conference, and remote teams. Toggle the signal channels or scrub the timeline to explore each read.
Three steps. Any video.
Upload a video
A talk, a game, an interview, a meeting. No wearables. No surveys.
The engine reads it
One person on camera: 468 facial landmarks per frame, plus body, voice, and words. A full room on one camera is read as a group instead, from a face count and which way the room faces.
Get the report
The scores, the moments that mattered, and what to do next.
Watch a clip become a measured read.
Bring any footage
A game, a keynote, an interview, a call. No wearables, no surveys, no special rig. If a camera saw it, the engine can read it.
The engine maps the face
With one person on camera it tracks 468 facial landmarks frame by frame, plus body pose, voice, and words from the same pixels. A full room on one camera is read as a group instead.
It reads four channels at once
Face, body, voice, and words. The engine tells you when they agree, when they conflict, and how sure it is.
Signals become scores
Eight proprietary scores, each traced back to measurable geometry and each carrying its own confidence level.
You get the read, not the footage
The moments that mattered and what to do next. We keep the numbers and delete the video once the run finishes. The read is yours.
Eight scores, read from the footage.
How steady a person stays under pressure, and how fast they settle.
How strongly a person holds a room. Read from gaze, posture, and real smiles.
The stretches where focus looks steady.
The whole instrument, in one view.
Four signals. One field.
Face, body, voice, and words, read together in a single pass over the same footage.
Confidence, not certainty.
Every finding is rated. When the signal is thin, the engine abstains instead of guessing.
Built on forty years of science.
Every score traces back to measurable facial geometry and four decades of peer reviewed behavioural research.
FACS / peer reviewedThe output is only as good as the method behind it.
Face, body, voice, and words. The engine tells you when the four channels agree and when they pull apart.
Footage deleted, numbers kept
The read runs in your browser, so the video stays there. Room mode, a second camera, or a file your browser cannot decode goes to a GPU server. There the footage is deleted when the run ends, or within the hour if it fails.
Confidence, not certainty
High, Medium, Low, or Abstain on each finding. A score drawn from 400 steady frames is not the same as one drawn from 12, and the report says which it is.