How It Works
How a presence check works
One API call at the consequential moment. The person's own device answers, policy turns the evidence into a verdict, and the verdict is sealed into a record you can verify without us.
Proof of Device
LiveHow does Presence know the device is trusted?
A passkey assertion, verified server-side, whose challenge commits to the exact action being approved.
Proof of Presence
LiveHow does Presence tell a live person from a replay or a deepfake?
Pulse under the skin, a randomized screen-reflection challenge, and lip–voice alignment — scored on the weakest signal, not the average.
Proof of Authority
LiveHow does Presence decide whether the action is permitted?
Policy turns evidence into one of four verdicts, scaled to what the action is worth and who is asking.
Action binding
LiveWhat stops an approval being replayed against a different transfer?
The session nonce commits to a hash of the action context, and the server re-computes that hash at decision time.
Signed records
LiveWhat evidence does a presence check leave behind?
Every decision — including the ones that were allowed — is sealed with an Ed25519 signature over its canonical form, chained to the record before it.
Failure behaviour
LiveWhat happens to my transaction if Presence is slow or unavailable?
The decision call is one HTTP request you time out on like any dependency. Telemetry is sideband, and one request returns every surface to watch-only.