Skip to content

VeryAI is now Very. Read the announcement.

Real authentication, backed by more than a decade of research

Very uses proprietary palm biometrics to solve the internet’s most urgent challenge: Authenticating real users and AI agents to stop spam, fraud, and account takeovers.

A single palm scan. Four steps.

All Very products are built around our proprietary palm scan.

Very is one of the only solutions to solve the verification trilemma

Read the litepaper
The verification trilemma: a triangle with corners Accurate, Scalable and Private. Face and ID documents sit between accurate and scalable, hardware scanners between accurate and private, and CAPTCHA, OTP and 2FA between scalable and private. Very sits in the centre, covering all three.

Accuracy, scalability and privacy

Very can be completed on almost any modern smartphone without any additional hardware. Very covers over 85% of the modern smartphone market.

Unlike other solutions, which require expensive or communally-owned hardware, Very can run authentications at global scale. We regularly verify tens of thousands of people every day.

Probability of a false accept (FAR) and a false reject (FRR) as the matching score rises. Both FAR lines fall and both FRR lines rise; the dual-palm, multi-scan lines sit well below the single-palm, single-scan lines. Probability Matching score
FAR single-palm + single-scan FAR dual-palm + multi-scan FRR single-palm + single-scan FRR dual-palm + multi-scan

Single palm FAR: 10⁻⁷, or 1 in 10 million

One of the fundamental measurements for biometrics is False Acceptance Rate (FAR), or the likelihood of a result giving a false match for two different biometric signatures. Very’s single palm verification achieves a FAR of up to 10⁻⁷,⁽¹⁾ while Apple Face ID reaches a FAR of 10⁻⁶,⁽²⁾ a 10x improvement and well above industry average.

That’s just a single palm. During registration, Very scans both palms multiple times for even further accuracy. If you were to scan both palms 5 times each for example, Very’s FAR compounds to 2.5 × 10⁻¹³, or 1 in 4 trillion.⁽¹⁾ At this level of accuracy, Very has the precision needed to uniquely distinguish individuals across a global population of eight billion and beyond.

False Acceptance Rate

Face ID
1 in 1 million
Very, single palm, 5 times
1 in 10 million
Very, dual palm, 5 times
1 in 4 trillion

Each palm scanned

Left palm

Right palm

8+ billion

people, each told apart

Very is able to combine palm scans with liveness detection, which asks the user to complete a randomized hand motion check in real time. This makes spoofing with an AI-generated palm image or video even more difficult.

Palms are more resistant to targeted attacks than other biometric modalities. They cannot be created wholesale from a digital footprint, like facial recognition (Unlike Facebook, there is no Palmbook). They are rarely lifted in real life, like fingerprints (Lifting whole palm prints is very difficult).

Liveness check

a random movement, asked in the moment

  1. Connect
  2. Open
  3. Check

Can’t follow a prompt it has never seen

  • Live handPasses
  • AI image or videoFails

Where an attacker can copy it from

Face
every profile photo online
Fingerprint
anything you touch
Palm
full print rarely left behind
Human-AI Collaboration Lab at Northwestern University

Fighting the deepfake arms race is of the utmost concern. That’s why we team up with the research team at the Human-AI Collaboration Lab, located at Northwestern University, to make sure we’re on the cutting edge of research and understanding.

How Very compares

Fingerprint, face, iris, DNA and Very palm rated for accuracy, security, scalability and privacy
AccuracySecurityScalabilityPrivacy
Fingerprint Medium Medium: Can be lifted from touches Low: Needs special hardware Low: Linked to identification databases
Face Medium Medium: Can be pulled from digital footprint High Low: Linked to identification databases
Iris High High Low: Needs special hardware Medium: Linked to identification databases
DNA High High Low: Needs special hardware Low: All biometric information
Very palm High: Face ID is 10x more likely to return a false acceptance.High: Entire palm print rarely left behindHigh: Can be used on any modern smartphoneHigh: Not linked to current identification databases

Technical requirements

Very works with more than 85% of smartphones worldwide.

  • iOS 13

    or more recent

  • Android 6

    or more recent

  • VGA camera

    720 × 720 or better

Compliance-ready

Very provides biometric authentication infrastructure rather than KYC decision-making and does not require government IDs for palm verification.

SOC 2 Type II certified, GDPR compliant, iBeta certified and BIPA compliant More compliance and trust

Stop bots and clean up your platform in hours

Get your API keys and access to the Very dashboard.