Skip to main content
Every verification answers two questions in order: is this a live human? and is this human new here? The first is the humanness check; the second is the uniqueness check. A person has to pass both.

​
Humanness

Humanness is a short, guided face check using the device camera. The person centers their face in an oval, holds steady for a few seconds, and a brief sequence of colors plays. That is the whole thing: about ten seconds, no documents, no license or passport uploads. The capture is scored for confidence that a real, present person is in front of the camera, not a photo, a video replay, a mask, or a generated face. Below the threshold, the verification doesn’t pass. The capture itself doesn’t stick around as a picture. It becomes a face code, a numeric representation used for the uniqueness check below. That’s the promise the flow makes on its consent screen: your selfie becomes a code, not a photo.

​
Any device, including desktop

The flow runs wherever the person is:
  • On a phone, the face check runs right there.
  • On a desktop with a webcam, the flow runs in place too. No phone required.
  • On a desktop, continuing on a phone is one click away: the page shows a QR code, the person scans it, and the same session continues in their phone’s browser. The desktop follows along over a realtime channel, so the moment the phone finishes, the desktop returns to your site on its own. Nobody copies codes or refreshes.

​
Uniqueness

Once we know the person is live, we check that they’re new. The face code from the capture is searched against the faces of people who have already verified. A match to this same person’s own record is recognition, not a problem. That’s how returning users get through quickly and how a member is recognized on a new device. A match to someone else’s record is a collision: the same face already verified under a different person. What happens next is up to two settings on your verification: One more face rule applies to signed-in members: if someone signs in with their identity and the new capture doesn’t match the face on their record, that’s a self mismatch and it’s reported the same way. This combination is what lets you say “one real human, once”: no duplicates, no second submission from the same face, scoped as tightly or as widely as your product needs.

​
What you get back

A confirmation reports verified: true only when the person cleared both checks. The confirmation also carries a status (approved, flagged, or denied) and the reasons behind it, such as collision. You confirm that result on your backend; see Confirmations.