Guardian OS™ · Operating platform

One governance kernel.Multiple operating environments.

Guardian OS is one operating environment for deploying, governing and supervising AI systems and autonomous agents — with Runtime Governance on the execution path.

The deployment profile decides where enforcement runs and who holds the keys. It does not change the control contract.

Book an enterprise briefingSee the operating model
One governance kernel behind a sovereign boundarySix deployment profiles — cloud, hybrid, private cloud, on-premises, sovereign and air-gapped — converge on a single governance kernel volume. A sovereign boundary encloses the kernel. The connector to an external control plane is drawn terminating at that boundary: the kernel requires no external control plane and no network.CLOUDHYBRIDPRIVATEON-PREMSOVEREIGNAIR-GAPPEDKERNELONE CONTROL CONTRACTSOVEREIGN BOUNDARYNO REQUIRED NETWORK
Control path — profile to kernelSovereign boundaryExternal control plane — not required, not crossed

The deployment profile changes where enforcement runs and who holds the keys. It does not change the control contract.

One integrated platform

Operate the enterprise. Govern the action. Preserve the evidence.

Guardian OS is the operating layer. Runtime Governance is the control boundary. The AI Twin is the current model of the enterprise. Intelligence Packs adapt the platform to the sector. Executive Workspaces turn governed state into decisions.

Enterprise operating model

Six capabilities. One source of governed truth.

Each capability has a distinct responsibility, but all operate over the same identities, policies, evidence and enterprise state. There are no disconnected governance dashboards to reconcile.

01

Enterprise Provisioning

Establishes enterprise identity, business units, environments, systems, agents and trust boundaries.

02

AI Twin

Continuously derives the operational model used across the platform.

03

Runtime Governance

Evaluates authority, policy, reachability and approval requirements before execution.

04

Industry Intelligence Packs

Adapt governance and evidence to the organisation’s sector and obligations.

05

Executive Workspaces

Present role-specific decisions, posture, approvals and operational briefings.

06

Managed Governance

Monitors drift, evidence quality and control performance throughout the operating lifecycle.

Common control plane

One Runtime Governance kernel across the platform.

  • Deny by default
  • Fail closed
  • Deny-only policy extensions
  • Evidence-backed verdicts
Continuous operating lifecycle

Governance continues after deployment.

Guardian OS does not end at onboarding. It continuously connects enterprise state, runtime decisions, sector obligations, evidence and executive oversight.

  1. 01

    Deploy or connect

    An AI system, agent or governed workflow enters the enterprise estate.

  2. 02

    Provision

    Guardian OS establishes identity, ownership, environment, permissions and control scope.

  3. 03

    Model

    The AI Twin derives the system, dependencies, trust relationships and reachability.

  4. 04

    Evaluate

    Runtime Governance assesses proposed behaviour against active policy before execution.

  5. 05

    Apply domain context

    The relevant Intelligence Pack adds sector controls and evidence requirements.

  6. 06

    Decide

    Allowed actions proceed; higher-risk actions route to the correct approval authority.

  7. 07

    Evidence

    Inputs, policy context, verdicts, approvals and outcomes are preserved.

  8. 08

    Brief

    Executive Workspaces update from the same governed source of truth.

  9. 09

    Monitor

    Continuous checks detect drift, policy change and emerging operational risk.

Continuous loopObserve → evaluate → decide → evidence → brief → monitor → re-evaluate
Executive outcomes

Control the estate without slowing the mission.

Know every AI system.

Maintain an accountable view of systems, models, agents, tools and owners.

Understand every dependency.

See what each system can reach and the downstream consequence of change.

Govern every action.

Apply policy at the moment an autonomous system attempts to act.

Detect drift automatically.

Surface divergence from the approved estate and governance baseline.

Maintain evidence continuously.

Preserve decision and execution evidence as operations occur.

Deploy with control intact.

Move from pilot to production without rebuilding governance around every agent.

Reduce operational risk.

Make authority, escalation and failure behaviour explicit before incidents occur.

Before and after

From fragmented AI oversight to an operating model.

Operating concernBefore Guardian OSWith Guardian OS
AI inventorySpreadsheets and periodic discoveryContinuously derived enterprise model
PolicyDocuments interpreted after the factRuntime evaluation before execution
Executive visibilitySeparate reports and inconsistent numbersRole-specific views of one governed state
EvidenceReconstructed for audit or incident reviewGenerated through the operating lifecycle
Sector controlsRebuilt for each programmeVersioned Intelligence Packs on one kernel
ChangeManual reviews struggle to follow driftBaseline comparison and continuous monitoring
Executive Workspaces

One enterprise. Role-specific command.

Every workspace reads from the same AI Twin, governance state and evidence system. Leaders see the decisions relevant to their mandate without creating competing versions of the enterprise.

CEO

Chief Executive

A strategic view of the enterprise — understood in under two minutes.

CTO

Chief Technology

Technical governance — the runtime, the estate and its topology.

CISO

Chief Information Security

Security governance — threats, escalations, blocks and evidence.

Risk

Chief Risk Officer

Enterprise risk governance — exposure, concentration, trend and the live risk register.

Compliance

Chief Compliance

Regulatory governance — posture, evidence and audit readiness.

Operations

Chief Operating

Operational governance — departments, approvals and automation.

Finance

Chief Financial

Financial governance — footprint, governed value and optimisation.

Legal

Legal & Governance

Legal evidence — decisions, versions, chains and attestations.

Industry Intelligence Packs

Sector intelligence without a separate platform.

Versioned packs contribute policies, regulatory mappings, evidence requirements and incident workflows to the same Runtime Governance kernel. They can add constraints; they cannot weaken the enterprise baseline.

Healthcare

v1.0.0

Clinical AI governance, patient-safety monitoring and regulatory evidence for healthcare enterprises.

Policies
3
Evidence maps
5
Workflows
2

Financial services

v1.0.0

Payment, trading, fraud and model-risk governance for financial enterprises.

Policies
4
Evidence maps
5
Workflows
2

Cybersecurity

v1.0.0

Threat intelligence, runtime-attack monitoring and privileged-action governance for security operations.

Policies
3
Evidence maps
5
Workflows
2

Government & public sector

v1.0.0

Department governance, procurement oversight and citizen-service AI governance for public sector bodies.

Policies
3
Evidence maps
4
Workflows
1

Manufacturing

v1.0.0

Robotics governance, operational safety and industrial AI monitoring for production environments.

Policies
3
Evidence maps
4
Workflows
2

Insurance

v1.0.0

Claims governance, underwriting AI oversight, fraud detection and regulatory reporting for insurers.

Policies
3
Evidence maps
4
Workflows
1

Retail & commerce

v1.0.0

Customer AI governance, pricing governance and supply-chain intelligence for commerce enterprises.

Policies
3
Evidence maps
3
Workflows
1

Education

v1.0.0

Educational AI governance, student-data protection and institutional oversight for education providers.

Policies
3
Evidence maps
4
Workflows
1

National security

v1.0.0

Classified governance, authority chains, mission approval and secure evidence for national security organisations operating AI inside a sovereign boundary.

Policies
6
Evidence maps
6
Workflows
4

Defence operations

v1.0.0

Governed autonomous operations, mission planning, command authority and deployment readiness for defence organisations running AI inside an operational command structure.

Policies
6
Evidence maps
6
Workflows
4

Critical national infrastructure

v1.0.0

Operational resilience governance for energy, utilities, telecoms, transport and water operators — dependency intelligence, operational risk, resilience metrics and incident workflows on one governed platform.

Policies
6
Evidence maps
6
Workflows
4

Public sector (sovereign)

v1.0.0

Departmental governance, procurement governance, citizen service workflows, policy implementation tracking and executive accountability for government departments operating AI under sovereign data and supply-chain obligations.

Policies
6
Evidence maps
7
Workflows
4

National healthcare

v1.0.0

Clinical AI oversight, patient data protection and cross-trust governance for national health systems — one governed platform across many trusts, boards and care settings.

Policies
6
Evidence maps
7
Workflows
4

Sovereign research & development

v1.0.0

Research integrity, export control, dual-use assessment and intellectual property protection for national laboratories, research agencies and defence science organisations running AI on sensitive research.

Policies
6
Evidence maps
6
Workflows
4

National cyber operations

v1.0.0

Operational authority, deconfliction, attribution discipline and vulnerability handling for national cyber defence organisations — ensuring no consequential cyber action is taken autonomously.

Policies
7
Evidence maps
7
Workflows
5

Pack counts are read directly from the shipping registry. Availability and regulatory scope are confirmed during assessment.

Explore Industry Intelligence Packs
Deployment profiles

Change the deployment boundary—not the governance standard.

Guardian OS can be aligned to enterprise, jurisdictional and infrastructure requirements. The hosting and connectivity model may change; the Runtime Governance kernel and decision contract do not.

Cloud

Operate Guardian OS within a cloud-aligned enterprise architecture.

Hybrid

Coordinate governance across cloud services and controlled private environments.

Private cloud

Keep platform services inside a dedicated enterprise cloud boundary.

On-premises

Deploy within organisation-controlled infrastructure where the engagement supports it.

Sovereign

Apply jurisdictional, residency and operational-control requirements to the deployment model.

Air-gapped

Support isolated operating environments where available and scoped for the programme.

Deployment profileCloud · Hybrid · Private · On-premises · Sovereign · Air-gappedUnchanged control contractIdentity → policy → verdict → approval → execution → evidence
Sovereign deployment

The same kernel, with the network taken away.

Guardian OS runs in 6 deployment profiles, from managed cloud to fully air-gapped. The Runtime Governance kernel is byte-for-byte identical in every one of them — only the providers behind it change. A sovereign estate keeps its state, its policies and its evidence inside its own boundary, and nothing about how a verdict is reached changes.

Proven today — enforced in code, asserted in CI
  • The Ω engine loads customer policy from a signed bundle on disk — no database, no control plane, no network.
  • Bundles are Ed25519-signed. The signing key never enters the environment; the engine verifies with the standard library alone.
  • Under a sovereign profile the platform refuses to build a cloud client even when credentials are present in the environment.
  • The interface makes zero external requests — no font CDN, no analytics, no embeds. Measured on every build.
  • Evidence packs, attestations and audit exports render to PDF on the box, with no browser installed.
  • CI runs the platform in a network namespace with no interface but loopback, and the engine in a container started with --network none.
Not yet — stated plainly
  • No deployment has run on customer hardware yet. The install media, images and acceptance suite exist; a witnessed site record does not.
  • No third-party accreditation. No Common Criteria evaluation, no NCSC assurance, no FedRAMP authorisation, no ATO.
  • No identity provider of our own — Guardian OS sits behind the estate's existing IdP.
  • No independent penetration test of the codebase has been commissioned.

Air-gapped operation is proven by continuous integration with network access removed, not by assertion. It has not yet been run on a customer site, so the accurate description today is acceptance-testable, not field-tested — and the acceptance suite marks any run without a named site and witness as a self-test on the face of the document. A control mapping and its open gap register are published rather than summarised. Request the sovereign deployment pack →

Enterprise integration

Governed where autonomous systems attempt to act.

Guardian OS sits at the agent and tool-call boundary. It evaluates the proposed action before it reaches the enterprise system, independent of the model that produced it.

Agent and orchestration layer

Where Guardian OS receives the proposed action for governance.

OpenAI AgentsLangGraphLangChainAutoGenAmazon BedrockMCP serversCustom orchestrators
Enterprise destination

Where the governed action would execute if policy and authority allow it.

Microsoft 365AzureAWSGoogle CloudSalesforceServiceNowSAPSnowflakeDatabricksPalantir

Named systems illustrate common destinations for governed actions. They do not imply a native connector or vendor endorsement. Review the integration model →

Enterprise adoption

Move from assessed risk to continuous operation.

Adoption is staged so architecture, policy, operational ownership and evidence can be validated before the platform becomes part of the enterprise control environment.

  1. 01

    Assessment

    Establish risk, readiness, operating scope and the first governed use case.

    1–2 weeks
  2. 02

    Enterprise Provisioning

    Model the estate, establish trust boundaries and configure the operating environment.

    Scope dependent
  3. 03

    Guardian OS Licence

    Run the platform, workspaces, governed agents, AI Twin and selected Intelligence Packs.

    Annual
  4. 04

    Managed Governance

    Monitor drift, evidence, policy performance and revalidation requirements.

    Monthly or annual
  5. 05

    Executive Advisory

    Support accountable adoption, operating-model decisions and executive oversight.

    Monthly or annual
Enterprise infrastructure for autonomous operations

Operate AI with visibility, policy and evidence built in.

Start with one governed use case. Establish the operating model required to scale.

Book an enterprise briefingExplore the AI Twin