Abstract
Organizations now assemble AI capability from models, agents, hosted services, autonomous machines and existing enterprise systems, and replace those parts far more often than they change who is responsible for the outcome. This paper argues that what an organization should hold stable is not its technology stack but its operational definition: what it is responsible for, who may authorize what, what evidence it must keep, and how it keeps operating when a component fails. Jynsyn calls that definition a Sovereign Operational Domain (SOD).
We follow one illustrative case: overnight security at a logistics company whose guard contractor now pairs a single on-site guard with a docked drone and a ground robot. It shows how the Jynsyn V3 reference architecture separates a durable responsibility from the replaceable system participants that serve it, how the company scales to a region served by a different contractor without rebuilding its authority model, and why, when an AI proposes that a drone follow an intruder off the property, the decision is not the drone's or the model's to make.
The claim is architectural: organizations need a durable object that composes established mechanisms around accountable authority, specified as requirements an implementation can be tested against. The paper ends with propositions that would show it wrong.
Four properties
Change without drift
Models, machines and contractors change; responsibilities and approvals change only by decision.
Access without authority
Admitting a system to a role never lets it act on its own say-so.
Failure within limits
An outage or a lost link narrows what is allowed; it never widens it.
Proof on demand
Every consequential action must trace to an accountable role and a named person.
The idea of a governed boundary isn't new. The SOD builds on the security domain and the authorization boundary from established security practice, with "sovereign" meant operationally: control over who runs, decides, responds and approves exceptions, held by accountable people. It isn't a data-residency or AI sovereignty zone, though it can use one. Jynsyn doesn't replace the platforms, agents and harnesses an organization already uses. A well-engineered harness enforces the SOD; it does not define it.
Status
The Jynsyn V3 Reference Architecture is forthcoming. When it's released, the paper will be reissued as a new version; the DOI above always resolves to the latest version. Jynsyn V3 is a reference architecture, not a product or a certification, and it hasn't yet been proven in the field.
| Version | Date | Notes |
|---|---|---|
| 1.0 | 2 October 2026 | First public edition |
Cite
@techreport{chandler2026sod,
author = {Chandler, Timothy E.},
title = {Sovereign Operational Domains: Durable Accountability for AI-Enabled Physical Security},
institution = {Bear Canyon Systems},
type = {White paper},
note = {Version 1.0},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23110479},
url = {https://doi.org/10.5281/zenodo.23110479}
}Questions, critique, or interest in building on Jynsyn? Contact Bear Canyon Systems