Lab / AAB

Agentic Authority Boundary

An architectural approach for governing how far autonomous reasoning is permitted to become operationally binding.

Layered diagram of Intent, Plan, Action and Execution with the governed authority layer highlighted

Patent pendingReference implementationExperimental evaluation

Overview

Autonomous agents can reason, plan and propose actions. The Agentic Authority Boundary defines the point at which that reasoning may become an operationally binding effect, and the controls that apply when it does.

Architecture

Intent

What the operator wants to achieve.

Plan

The reasoned proposal produced by the agent.

Action

The authority boundary check before anything binding happens.

Execution

The operational effect, within the authority granted.

Reference implementation

A working implementation used to explore the boundary in practice. Details are available on request.

Experimental evaluation

Evaluated through simulation and real-world scenarios. Findings are shared with research and industry partners.

Paper

The paper describing AAB is available on request through our contact page.

Patent status. A patent application has been filed and is pending. The current patent application is personally owned by Richard Knuszka.

Interested in governed autonomy?

We are happy to discuss AAB with research and industry partners.

Get in touch →