Market · Signature intelligence

Next

A signature is a movement, not a picture.

Every check for a forged signature today compares shapes, and a practised forger can copy a shape. What a forger cannot copy is the motion: the speed into a curve, the pressure on a downstroke, the rhythm and stroke order drilled by years of habit. The pen records exactly that, on ordinary paper, as the signature is made.

Understanding the problem

Shape comparison is what forgers practise against.

Signatures still authorise bank mandates, insurance claims, property registrations and wills across India, and disputes still come down to an expert comparing two images. A skilled forger practises until the image matches. But the image is the residue of a movement, and the movement is where identity actually lives: pen dynamics are drilled into muscle memory over thousands of repetitions and are extraordinarily hard to imitate at natural speed. A signature captured with the pen carries its own motion evidence: how it was written, not only how it ended up looking.

Next on the roadmap, not a product today. Nothing here is a forensic or legal claim. This page describes the application we are building toward and the partners we want to build it with.

Two moments, one signal

Enrol the movement. Verify the movement.

Enrol

A motion profile, not a specimen card

The signer writes a handful of natural signatures with the pen on ordinary paper. The dual-IMU signal at 208 Hz captures the dynamics: stroke order, velocity curve, pressure pattern, rhythm and pen lifts, the features habit makes stable and imitation makes unstable.

Replaces the static specimen signature banks keep on file today.
Verify

Compare how, not just what

At signing time the same pen records the live dynamics and compares them against the enrolled motion profile. A traced or practised copy can match the shape while the velocity and pressure story gives it away.

The document still gets its ink signature; the motion check rides along.

Key benefits

Why motion is the right tool here.

Harder to fake

Dynamics resist imitation

Shape can be traced; the velocity and pressure pattern of a practised hand at natural speed is another matter entirely.

Familiar

Ink stays ink

No signing pads or styluses: the same pen-on-paper ritual every branch and registrar already uses, with the signal captured alongside.

Evidential

A record of the act

Each verified signature stores its own motion trace: time-stamped evidence of how the mark was made, not just the mark.

Common use cases

Where it could help

  • Banking, high-value mandates, lockers and cheque disputes where the specimen card decides.
  • Insurance, claim and proposal signatures with motion evidence attached.
  • Legal & registration, wills, deeds and agreements where authorship is later contested.
  • Examinations & attendance, verifying the signer is the enrolled candidate.

One example

From a disputed cheque to motion evidence

A branch flags a high-value cheque: the signature looks right, but the account holder denies signing. Today that dispute rests on two images and an expert's eye.

With enrolment on the pen, the branch compares the disputed signature's motion trace against the holder's profile: the downstroke pressure and the rhythm between the two pen lifts do not match the habit of ten years of signatures. The shape agreed; the movement did not.

Enrol the habitSign on paperCompare the motionEvidence attached

How we’d build it

Capture, characterise, harden, then deploy.

01 · Capture

Consent-first signatures

Motion gathered with explicit consent across ages and scripts; the dataset work Vahini already does, applied to signatures.

02 · Characterise

Stability & imitation

Measure which dynamic features stay stable for a signer and collapse for an imitator, across sessions and moods.

03 · Harden & deploy

With institutions

False-accept and false-reject rates proven with banking and legal partners before anything touches a real workflow.

Signature dynamics have decades of research behind them; the pen’s contribution is capturing them on ordinary paper in the ritual institutions already trust. We publish error rates before we ask anyone to rely on them.

Work in banking, insurance or legal?

This application is built with institutions, not around them. Help us define the enrolment, the thresholds and the evidence format your workflow needs.