Identity-verified hiring

Same person. All the way through.

One identity check at the start. A free check at every round after. And one more on the first morning — because that is where impersonation actually pays off.

Pre-launch — working with a small number of early partners

The chain

Verified once. Checked at every step. Closed on day one.

  1. Once Enrollment Government-anchored. About 90 seconds. Never repeated.
  2. Round 1 Technical A check against the enrollment. Free, on-device.
  3. Round 2 Technical Same check. No round is exempt.
  4. Round 3 HR Because the swap usually happens between rounds.
  5. Day one The close One check before systems access. The loop shuts.

Most verification stops at the first dot. The payoff for impersonation is at the last one.

The idea

Detection expires. Binding doesn't.

Every deepfake detector is trained on the fakes that existed when it was built. Fakes improve, accuracy quietly erodes, and someone has to retrain. That is a treadmill.

Linking a real identity to a device, to each session, to the first day at work is a different kind of claim. A better fake does not un-link facts.

So we treat detection models as parts that get replaced. The chain they plug into is the part meant to still be right in ten years.

Two sides

Verification only works if both sides want it.

Employers

Evidence, not assurances

  • Works with the interview process you already run
  • A signed record you can hand an auditor
  • Flags reach a human with the evidence attached
  • Runs on your own infrastructure if compliance needs it
Candidates

Verify once, carry it with you

  • Free — employers pay, candidates never do
  • One enrollment, reusable across employers
  • Roughly ten seconds per round after that
  • No system can reject you; only a person can

Why now

Three things changed at once.

01

Hiring went remote and stayed

The interview became a video call. Nobody rebuilt the identity assumptions that a room used to provide for free.

02

Convincing fakes got cheap

Real-time face and voice synthesis moved from research to consumer tools. The skill floor for impersonation collapsed.

03

Identity rails matured

India now has consent-based digital identity infrastructure good enough to anchor against. That was not true a decade ago.

Risk-based

Friction is a budget. We spend it where risk lives.

Standard · most hiring

A link and a quick check

Before each round. Nothing to install, nothing running during the call. The path most candidates take.

Strict · selective

Dual-device, continuous

Phone paired as a trust anchor, checks through the session, environment signals. For senior and regulated roles.

The employer sets policy. A risk assessment can raise a session's mode, never lower it.

Where we draw the line

We verify presence. We don't claim to read minds.

No system catches everything Ours included. Our contract language is fraud-resistant with auditable evidence — never fraud-proof, and we won't let a sales conversation drift there.
Nothing is auto-rejected VAHI cannot reject a candidate. It produces evidence; a named human decides. That is structural, not a setting.
The interview never waits on us If we go down, the interview proceeds and is marked unverified. Only the certificate waits.

Data

Templates, not photo galleries.

A face check produces a mathematical representation, not a picture. Interview checks send scores and events, not video. Flagged material is deleted after 90 days by default. Deployable entirely on your own infrastructure.

If proxy interviews have already cost you, let's talk.

We're pre-launch and working with a small number of early partners. One inbox, and a person reads it.