How The Read works
Ninety seconds on camera, 30 signals, and a plain answer about what a room sees in you before you speak. All of it computed on the device in your hand.
What it measures
The reader watches your face and body about thirty times a second. On every frame it computes head pose, gaze offset, eye openness, blink state, facial asymmetry and jaw tension; from the same frame it derives the action units that expression research uses — brow furrow, cheek raise, lip tension. When your shoulders are in frame it adds postural signals: sway, symmetry, lean, how much your hands move.
None of those numbers are interesting on their own. What matters is how they behave over time and under load — whether your line holds, and what happens to it when something changes.
Why ninety seconds is enough
Because pressure shows up in seconds, not hours. A ninety-second read is a few thousand samples of the same face, which is enough to establish what steady looks like for you and then to notice when you leave it. That is the whole method: every score compares you to your own baseline, not to a population average. A naturally still person and a naturally animated one are each measured against themselves, because that is the only comparison that means anything when the sample size is one person for ninety seconds.
At some point during the read, something happens on screen. You are told in advance that it will; you are not told when. What you do in the seconds after it is the most specific thing the reader learns about you.
What each score means
Composure
How steady you held. Built from the absence of stress markers, how quickly you settled after them, and how still your head stayed. It is not the same as being frozen — a person who reacts and recovers can score higher than one who never moved.
Presence
How much of you the camera actually got. Facing it, expressive enough to read, open rather than braced. Someone can be perfectly composed and barely present; the two describe different things and they come apart more often than people expect.
Peak Alignment
The only score that compares you to other people. It measures how closely your pattern tracks what has been documented in performers under real load — surgeons, soloists, closers, trial lawyers. High means close to that profile, which is not automatically better.
Composure Δ
Whether the second half of your read looked like the first. A small delta means the pressure changed and you didn't.
Your camera feed never leaves your device
This is the part worth being specific about, because most tools that say something like it are being careful with words. The models run inside your browser. There is no server that sees your video, no frame is transmitted anywhere, and there is no upload step to opt out of — the capability does not exist in the product.
What can leave, and only if you ask: the four scores and your archetype, sent to your own email address. If you separately choose to keep the still photo from your read, that one image is uploaded and you are asked plainly first. Everything else — the frames, the landmarks, the second-by-second trace — lives in the tab and dies with it.
Who it is for
Anyone whose job includes a moment where a room decides something about them. The scan is the same in every case; what changes is the situation you point it at.
- PitchKnow your tell before the room finds it.
- InterviewThe question you hoped they wouldn’t ask.
- SpeakingThe room reads you before word one.
- Pre-gameGame face, measured — what the other bench sees.
At the end you come back as one of 24 archetypes. There are others that are not on that list.
Common questions
- Is anything recorded?
- No. The video never leaves your device — there is no upload step to opt out of, because there is no upload. Your browser runs the models locally and only the resulting scores are ever sent anywhere, and only if you ask for your Read by email. If you close the tab before then, nothing survives.
- How long does it take?
- Ninety seconds is the ceiling, not the requirement. The Done button unlocks after five seconds and the reader tells you when it has seen enough to have a real opinion — around thirty seconds of clean detection. Longer reads are steadier reads, but you are never held there.
- What do the scores mean?
- Composure is how steady you held across the read. Presence is how much of you the camera actually got — facing it, expressive, not withdrawn. Peak Alignment compares your pattern to people measured under real load. Composure Δ asks whether the second half of your read looked like the first.
- What is Peak Alignment?
- It is the only score that compares you to anyone else. The others describe you against yourself; Peak Alignment measures how closely your gaze steadiness, postural stability and blink pattern track the profiles documented in performers under genuine pressure — the quiet-eye and quiet-body literature. High is close, not better.
- Do I need an account?
- No. There is no sign-up and no password. A Read is anonymous unless you choose to put a first name on the leaderboard or ask for the result by email.
- What does it need to work?
- A camera and a browser. It runs on a phone, and a phone is the best way to use it — you get the front camera pointed at your own face and nothing else to hold.
The room reads you first.
Get my Read