NeuralPrint detects deepfakes and proxy interviews.
it reads how they
think.
Not just what they say. NeuralPrint is the layer that reads a candidate the way a person would, only continuously, and without getting tired.
What NeuralPrint detects during an interview.
NeuralPrint axon reads a candidate for the whole session rather than at one moment they can prepare for. These are the four things that read catches.
A face that isn't theirs
Rendered, not liveA deepfake holds together for a moment and comes apart over a session. A rendered face does not move, light, or react the way a live one does, and the gap widens the longer it has to hold.
Someone else on the call
Not the applicantAnswers coming from somewhere else
Not their ownAn operation, not an applicant
One of manyWe publish what axon determines, not how it determines it. The people this is built to catch read marketing pages too.
What it reads
What NeuralPrint reads: face, voice, gesture and timing.
Four channels, read continuously and fused into one profile: the face and how it moves, the voice and how it is produced, gesture and posture, and the timing between a question and an answer. No single channel decides anything; the profile is what they agree on.
face
micro-expression timing, gaze behavior, whether the framing and lighting hold up under motion.
voice
cadence, pitch variance, and whether the voice matches the face across the whole session.
gesture
how someone actually moves when they think, not just what they say once they've thought.
timing
how long an answer takes to start, and whether that latency matches a rehearsed script.
0+
Signals per interview
Collapsed into one structured profile, ready the moment the interview ends, and consistent across every candidate who applies.
Continuous, not momentary
Why a one-time identity check is not enough.
A photo match at sign-up, a liveness check at the start of the call. everything in between is unwatched, which is exactly where a proxy or a rendered face slips through. NeuralPrint reads for the whole session, so there's no gap between checks for anything to hide in.
A momentary check
Four snapshots, gaps in between
NeuralPrint, the whole session
Read continuously, no gap to slip through
The measurement
How consistency over a session reveals a deepfake or a proxy.
A rented face and a live one can both look convincing for a moment. they behave differently over fifteen minutes. NeuralPrint measures how a candidate's own signal drifts, or doesn't, against itself across the session, and flags the moments it can't reconcile.
See what that looks like on a real sessionSpec sheet
NeuralPrint specifications.
runs on
every verified session, start to finish. the same engine behind identity verification and biometric authentication.
outputs
one structured profile per candidate, attached to the record your team reads.
does not publish
the exact signal list. the people we exist to catch read marketing pages too.
pairs with
λ-CORE, which scores the evidence NeuralPrint verifies. see how they divide the work.
hire who is right.
Switch it on today. Tomorrow you read a shortlist, not a pile.
Free to start, no credit card, set up in minutes