Research and Frameworks · Architecture

Human OS.

People > Process > Technology

Inverting the industry status quo, where people come last.

Human OS is our architecture for a privacy-first ecosystem where connected apps, user-owned agents and trust infrastructure work in unison to provide private personalization: the ability to have software work for you without surrendering identity, data or control.

Not a traditional, monolithic operating system. It is a modular way to think about software that lets individuals carry context, delegate intent, verify information, and interact safely across systems.

Human OS architecture diagram

User-centric architecture. Trusted AI assistance.

Human OS puts the person at the centre, shielded by a self-governed boundary and assisted by local AI with a personal harness.

  • CentreThe person: their identity, data and intent.
  • Agent layerAgents that act on the person’s behalf, carrying context across services.
  • Trust infrastructureIdentity, verification, permission, provenance, security, memory and a governance kernel, built into the plumbing.
  • App layerProductivity, health, commerce, learning and the rest: useful, and kept at the edge.
  1. Privacy-first

    Identity and data stay with the person. A service receives only what the task needs.

  2. Intelligent

    Software understands intent and does the work, instead of handing it back to the person.

  3. Personalised

    It adapts to the individual without extracting their data to do it.

  4. Ethical

    Designed and governed by principles that put people first, as set out in Abundance, our ethics framework.

What it enables

Autonomy for people. Advantage for business.

Embedding autonomy and trust in infrastructure aligns incentives and decreases friction between users and businesses.

For people

Software that works for you, on your terms.

  • Keep your identity and own your data, deciding granularly who sees what.
  • Sign in without passwords.
  • Browse without fear, with verification behind what you see.
  • Use reputation to trust what you read and interact safely with strangers.
  • Build capacity and amplify productivity, safely and privately.

With individuals insulated, safer communities can form around shared interest, where self-regulation comes from incentives, not enforcement.

For businesses

Trust becomes a growth advantage.

  • More accurate data, with integrity you can verify in your analytics.
  • Lower customer acquisition costs and less wasted ad spend.
  • Deeper loyalty that turns into advocacy and lifetime value.
  • Fewer breaches and phishing attempts, with external interfaces and communication separated cryptographically.

A network that prioritises user autonomy is inherently a safer network.

Solving trust

Existing technology, new methodology.

Trust isn’t one problem. Split into its parts, each can be solved with technology that already exists.

Privacy Identity Authenticity Knowledge Truthfulness Accountability Trust in six parts
BuiltPlanned
  1. Privacy

    Built in Bionic

    Sign-ups demand a name, email and phone. Every copy is one more place to leak from.

    Decentralised identifiers A different identifier for every service. It recognises you when you return, and holds nothing worth stealing.

    The Cost of Exposure →
  2. Identity

    Built in Bionic

    Accounts belong to the platform, and passwords get phished, reused and leaked.

    Passkeys An identity you hold yourself, unlocked by a key that never leaves your device.

  3. Authenticity

    Built in Bionic

    Phishing was the most reported crime to the FBI’s internet crime centre in 2025. It works because anyone can claim to be anyone.

    Digital signatures Signed, permissioned communication: you know who sent it, and why it was allowed to reach you.

    See our signed business profile →
  4. Knowledge

    In development

    An assistant that forgets you can’t act for you. One that remembers everything keeps a record nobody approved.

    Governed memory It knows what you mean and where to find it, inside your own boundary. It moves with you between models, and you can export it.

  5. Truthfulness

    Planned in Bionic

    Reviews and ratings anyone can buy, from accounts no one can see.

    Reputation systems Reviews corroborated by other users, and weighted toward a smaller set of trusted reviewers.

  6. Accountability

    Planned

    Anonymity without consequence rewards whoever games the system, and agents act with authority nobody can trace.

    Incentive design Gaming costs reputation, and with it access, rewards and group promotions. Behaviour regulates itself.

    Governance kernel An agent acting for you stays within the purpose and time you gave it, and every action traces back to you and the policy it followed.

Where it is built

See the products carrying it.

Anchored in Abundance.

Abundance is our ethics framework, and the standard every product built on Human OS is held to. Its twelve principles are design constraints rather than aspirations, which is what keeps human-first enforceable.

Explore Abundance