THE CONSTITUTIONAL PLATFORM

Intelligence can decide. Authority must govern.

Zortrex is building an independent constitutional layer between AI intention and real-world execution - so consequential actions can be verified, authorised, released or refused.

Explore the platform

Read the principles

EXECUTION REQUEST

AUTHORITY ENGINE

Independent Constitutional Authority

Identity ✓
Authority ✓
Purpose ✓
Consent ✓
Scope ✓
Live Conditions ✓

DECISION

Release

Refuse

VERIFIABLE DECISION EVIDENCE

Independent decision record

STATUS

Verifying…

THE PROBLEM

AI is gaining the power to act. The authority model has not caught up.

Modern AI can call tools, access systems, move data, trigger workflows and increasingly affect the physical world. The route from AI intention to execution is often little more than inherited permissions or application-level guardrails.

Abstract blue network of connected points

THE PLATFORM

A constitutional boundary for consequential execution.

AI can propose an action. Zortrex independently determines whether the conditions for execution are actually present.

01. Verify authority

02. Test the action

03. Release or refuse

CONSTITUTIONAL PRINCIPLES

The rules sit outside the intelligence.

Identity

Know precisely which actor is proposing, requesting or approving an action.

Identity

Know precisely which actor is proposing, requesting or approving an action.

Authority

Confirm that a valid mandate exists before execution is ever permitted.

Authority

Confirm that a valid mandate exists before execution is ever permitted.

Purpose

Ensure each action remains tied to a declared and defensible intent.

Purpose

Ensure each action remains tied to a declared and defensible intent.

Consent

Respect the permissions, choices and revocation rights of every affected party.

Consent

Respect the permissions, choices and revocation rights of every affected party.

Scope

Keep every permitted action proportionate, bounded and fully observable.

Scope

Keep every permitted action proportionate, bounded and fully observable.

Built for the point where AI decisions become consequential.

Financial actions, sensitive records, connected systems and enterprise agents all need an independent authority boundary.

WHERE IT MATTERS

CONSEQUENTIAL EXECUTION

Verified in context

Authority is tested at the moment release is requested - not assumed from static access.

WHITE PAPERS

The foundations of constitutional AI.

Authority Before Execution

Why consequential AI requires a distinct constitutional layer.

The Refusal Principle

Designing systems that can decline illegitimate action with clarity.

Evidence as Infrastructure

A durable record for machine intention and human authority.

INSIGHTS

Authority, governance and the future of agency.

The right to refuse

Why regulated autonomy depends on a system’s capacity to say no.

Agency needs a boundary

A constitutional view of AI action in the real world.

From policy to proof

Turning governance language into an operational decision record.

THE ZORTREX POSITION

AI may propose the action. It must never authorise itself.

As intelligence becomes more powerful, execution authority should become more explicit - not less.

Start a conversation

Zortrex, the constitutional platform