CYBERSECURITY · AT THE HEART OF THE PRODUCT

LOCAL Intelligence on your side. Control in your hands.

We design C.A.T.Y. to help from your own computer. Local processing, reviewed actions and security you can use without relying on sight.

Explore the published work
AIntegra logo: a cat silhouette in navy and lavender whose negative space forms a keyhole.

PRIVACY FROM THE START

Your computer is the centre. Your information belongs here, too.

C.A.T.Y.’s architecture prioritises processing information on the device itself. Investment will help us make the local model more capable.

The idea is simple: help with what is on your screen while you stay in control of your data and the actions being taken.

MORE THAN A PADLOCK

Security at every step.

  • Before taking action

    A control layer reviews what AI proposes. Sensitive actions require confirmation.

  • You decide access

    Permissions are part of the design: what the assistant can read and what it can interact with.

  • Local context and memory

    Preferences and context are designed to stay on the device, under the person’s control.

Product architecture: we continue to expand and refine its protections.

AI PROPOSES. YOU STAY IN CONTROL.

Between asking and acting, there is a safety layer.

  1. Understand
  2. Propose
  3. Confirm
  4. Act
  5. Verify

Confirmation is requested when an action requires it, such as before sending or deleting.

How we approach it

Protection must be accessible, too.

  • SCREEN AND AUDIO

    Safe mode

    A visual curtain and private audio for entering sensitive information away from nearby eyes.

  • AGVS · GESTURES

    Gesture verification

    An alternative to visual CAPTCHAs: verify that you are a person through gestures, without deciphering images.

Safe mode is designed; gesture verification is under development within the open framework.

ACCESSIBLE CYBERSECURITY, IN THE OPEN

OSCF Handbook

Our open framework for accessible cybersecurity. It includes AGVS: a proposal to verify that a person is present through gestures, without relying on images.

It goes beyond an idea: principles, risk analysis, interaction rules and a way to evaluate proposals. Other teams can read it and contribute.

Public draft v0.1

Explore the published work
  1. 01

    What we protect

    Principles, scope and limits of the proposal.

  2. 02

    What could go wrong

    Risks such as gesture replay or session abuse.

  3. 03

    How people use it

    Multiple ways to interact, without relying on sight alone.

  4. 04

    How we assess it

    Criteria for assessing accessibility and protection.

A design with clear references.

  • WCAG 2.2 level AA
  • ISO/IEC 27701
  • ISO 27001

Design references and development goals. We do not yet hold certifications.

What does your organisation need?

Tell us what information you need to protect and we can discuss how it fits the project.

Talk to us