Skip to content

How it works

Five steps, cryptographically grounded.

One round trip: the device proves what it is, your server asks us whether that clears your bar, and you act on the answer. Nothing to store, no personal data, no client-side keys.

  1. Nothing on the device worth stealing

    The chip makes the proof. Your client holds no key and renders no verdict — so there is no secret on the user's machine for an attacker to lift or resell.

  2. A captured proof is worthless

    Your server names a one-time value and the chip signs that exact one. Replaying yesterday's proof, or one lifted off another machine, fails on arrival.

  3. The cheap path to a fake closes

    A real chip carries a manufacturer signature a rented cloud server can't produce. That single check kills the $0.10 fake account, which is where abuse economics actually live.

  4. You draw the line, not us

    Your policy decides which devices count, changed without a redeploy. Allow cloud and we still bind each proof to one instance — so one pass can't become a thousand.

  5. An answer you can act on, and no new liability

    One call returns pass / warn / fail plus a device id that's stable for you and meaningless to anyone else. No token to store, no session to protect, no personal data to breach.

Read the full mechanism →quote construction, EK chain validation, policy resolution, cloud cross-validation

See the whole flow run on your own device.

Free up to 1,000 active devices a month, no card.