![n[act]-it](/__l5e/assets-v1/cdb90966-998a-47c1-acf8-9661e7bf2490/nact-it-dark-logo.png){kind=logo}
n[act]-it a product of iThoth Systems Inc.
Introducing Authority-Native Computing
Intelligence Is Not Authority.
The Authority Infrastructure for Agentic AI.
AI agents are becoming economic and operational actors. They can move money, negotiate transactions, purchase goods, operate equipment, and act across organizational boundaries.
But the ability to act is not the authority to act.
n[act]-it is developing Authority-Native Computing: a mathematical and computational foundation that makes legitimate, bounded, and verifiable authority a requirement of consequential AI operations.
Our infrastructure is designed to determine whether an exact operation is authorized, enforce its constitutional constraints, and preserve the evidence needed to verify what happened.
We are building the authority layer beneath autonomous intelligence.
Interactive demonstration
AI has capability. But does it have authority to act?
Give an autonomous agent a transaction. Change its circumstances. Watch constitutional authority determine what can legitimately proceed.
Proposed operation
An autonomous purchasing agent requests GPU compute from an approved provider, acting under a limited purchasing grant from its principal.
Governing grant · fixed
- Authorized purchasing ceiling: $2,500
- Supplier must be approved
- Second authorized approval required above $1,500
- Grant may be revoked by the principal
- A consumed operation cannot be replayed as fresh authorization
Scenario shortcuts
Constitutional evaluation
Capability does not confer authority.
An AI may be capable of completing a transaction. That does not give it the legitimate authority to do so. Authority must originate from a valid grant and survive verification of the exact operation and its governing conditions.
Evidence can resolve a condition. It cannot create a grant.
Approval evidence can satisfy an existing grant's conditions. It cannot enlarge the grant — a request beyond the authorized ceiling remains denied until a separately legitimate grant-changing event occurs.
Interactive constitutional simulation. No real transaction is executed.
Authority-Native Computing
A New Foundation for Autonomous Action
Traditional software permissions answer whether a user or system can access a resource.
Authority-Native Computing addresses a different question:
Is this precise consequential operation legitimately authorized, under these constraints, at this moment?
01
Constitutional Computation
Determines whether a specific operation satisfies legitimate authority, grants, constraints, and required evidence.
02
Constitutional Dynamics
Governs how delegated authority and its consequences remain bounded through time, across interactions and changing conditions.
03
Verified Enforcement
Connects constitutional decisions to execution controls and auditable outcomes, so authorization is not merely advisory.
These are architectural principles guiding development, not a claim of complete production-system verification or completed physical enforcement.
The infrastructure
From Mathematical Authority to Executable Infrastructure
01
Constitutional Grant
Explicit authority, granted by a principal with defined scope and constraints.
02
Authority Evaluation
The exact operation is evaluated against its grant, required evidence, and current constitutional state.
03
Governed Execution
Constitutional decisions are connected to execution controls and operative bounds.
04
Verifiable Evidence
Structured records connect the operation, authority evaluation, and auditable outcome.
- Authority is explicitly granted, not inferred from intelligence or capability.
- Delegated authority remains subject to its original constraints.
- Interactions do not create additional authority merely through composition.
- Revocation, expiry, and changes in authority must affect permissible execution.
- Consequential operations produce structured evidence for audit and verification.
Commercial applications · In development
Where Autonomous Intelligence Meets Real Authority
Agentic Commerce
Buyer and seller agents negotiate transactions and coordinate commitments without creating authority beyond the grants provided by their principals.
See the live demonstration →Financial Operations
Autonomous systems initiate or prepare consequential financial actions subject to explicit authority, amount limits, destination constraints, and verifiable approvals.
Enterprise Agents
Multiple agents work across organizational systems while respecting delegated authority, separation of duties, revocation, and auditable execution boundaries.
Autonomous Machines
Robots and drones operate under formally defined mission authority and enforceable operational constraints. Physical enforcement remains a research objective.
Our PX4 Software-in-the-Loop demonstration establishes an integration testbed, not a completed constitutional flight-control product or completed physical enforcement.
The architectural distinction
Beyond Access. Authority for the Exact Operation.
Access to a system
Traditional access control decides who can access a system. IAM, permissions, guardrails, and monitoring provide important identity, policy, safety, and oversight capabilities.
Authority for a consequence
Authority-Native Computing determines whether a consequential operation is legitimate, properly bounded, and executable under the current constitutional state.
Capability ≠ Authority.
A system’s ability to perform an operation does not establish its legitimate authority to perform it.
Public research overview
The Mathematics of Legitimate Autonomous Action
An active program of formal specification, mechanically checked models, and executable infrastructure development.
Formal methods research
Authority Conservation
Investigating and proving properties that prevent interaction or computation from illegitimately amplifying delegated authority.
Ongoing research
Constitutional Emergence
Extending constitutional models to evolving intelligences and collectives, including transformations of identity, participation, and governed authority.
Ongoing research
Physical Consequence Bounds
Research into connecting legitimate constitutional authority with independently constrained physical execution, using formal reachability and resource-bound models.
Bounded authority through change
Our models investigate how legitimate authority remains bounded as agents delegate, interact, and encounter changing conditions. Model verification is not a claim of universal security or a completed physical enforcement system.
The company
Building the Authority Infrastructure for the Agentic Era
n[act]-it is a deep-technology company developing the mathematical and computational foundations for governed autonomous action.
As artificial intelligence becomes increasingly capable of acting in the world, legitimate authority must become as fundamental to consequential computing as identity, networking, and execution.
Our mission is to make that possible.
Move Authority, Not Money.
We are not a payments processor or a replacement for existing financial rails. Our objective is to govern the authority under which autonomous systems may initiate and execute consequential operations.
Built by iThoth Systems Inc. — meet the team below.
The team
The people building the authority layer for autonomous computing.
Founder & CEO
Gary Bedell
- Retired Canadian diplomat, VIP security lead, international journalist and cybersecurity expert.
- Former aide to Nelson Mandela.
- MBA, University of Western Ontario (1986).
- Advanced Cyber Security Certificates, York University (2021).
- Founder of Intentics and creator of Constitutional Computation.
- Focused on the next execution layer of computing: legitimacy, authority, and autonomous execution.
Co-Founder & CCO
Ahsan Khan
- Experienced entrepreneur who navigated the startup process and founded his own AI automation company.
- He also brings several years working as a technical project coordinator at Big 5 banks in Canada. Demonstrates superior CRM skills.
- First Place | Ibtikar Toronto Shark Tank Competition, York University (2025)
- Spotlight Award | TD Insurance (Q1 2023)
- Project Management Professional (PMP) | Project Management Institute (2024)
Collaborative Engineering
Built with AI. Guided by Humans.
- The Constitutional Runtime has been developed through close collaboration between human architects and multiple frontier AI reasoning systems.
- The architecture, implementation, and technical documentation have been accelerated through disciplined human–AI collaboration while remaining directed by human vision and engineering judgment.
- This is not a replacement for engineering. It is a demonstration of what engineering can become.
Partner on the Authority Layer.
For enterprises, infrastructure providers, and AI agent platforms exploring governed autonomous execution.
Discuss a concrete operation, the authority it requires, and the constraints it must preserve. We welcome design-partner conversations across commerce, financial operations, enterprise systems, and autonomous machines.
Intelligence determines what an agent is capable of doing. Authority-Native Computing determines what it is legitimately permitted to do.