Our Approach
Constitutional Computer Science
Triarcus Systems approaches computation as a governed domain rather than an unconstrained capability. As computational systems grow in scale, speed, and autonomy, the question is no longer only what systems can do, but what they should be permitted to do, under whose authority, and with what accountability. Our work focuses on the scientific foundations required to answer those questions rigorously.
Governance Before Execution
Modern systems are often optimized for performance or autonomy in isolation. While these goals have value, they are insufficient in high-trust, safety-critical, or institutionally regulated environments. Triarcus begins from a different premise: power must be governed before it is exercised. Rather than adding oversight after deployment, we design governance directly into computational architectures themselves.
First Principles, Not Features
Our approach operates at the level of first principles. We develop constitutional models, formal invariants, and runtime governance architectures that preserve safety, agency, and accountability as systems scale. Software implementations are downstream expressions of this work, not its starting point.
Applied Where Trust Matters Most
Our work is intended for environments where failure is unacceptable, authority must be legitimate, and explanations matter as much as outcomes. In these contexts, governance is not overhead — it is infrastructure.
Human Authority Preserved
Advanced systems must remain human-directed. Our architectures ensure that authority is explicit, decisions are explainable, escalation paths are bounded, and oversight is preserved by design. Autonomy is constrained by legitimacy, not hidden behind optimization. Triarcus exists to ensure that as computational power increases, structure, legitimacy, and human agency increase with it.
Principles
Triarcus Systems is guided by a small set of foundational principles that shape all of our work:
Governance is infrastructure
Authority, constraint, and accountability must be architectural, not procedural.Restraint scales with power
As systems gain capability, limits and oversight must become more explicit, not less.Human agency is non-negotiable
Advanced systems may assist, but they must not obscure or replace legitimate human authority.Decisions must be explainable
Outcomes without re-constructible reasoning are operationally and institutionally unsafe.Legitimacy precedes optimization
Systems must be permitted to act before they are allowed to optimize.
These principles are invariant across domains and implementations.
The Founder
Triarcus Systems was founded by Michael G. Cooper, who began this work on October 1, 2025, in Grantsville, Utah. He had no prior background in artificial intelligence. He had a question about how governance and creation relate to one another — and he followed it where it led.
The work did not come from institutional backing or funded research. It came from a question, followed honestly to its conclusions, and the record of that work is intact and timestamped from the first day forward.
Michael is a senior business analyst working in courtroom software, serves as a Young Men's advisor in his LDS ward in Grantsville, and lives with his family in Tooele County, Utah.
The Name
Triarcus — from the Latin tri, "three," and arcus, "arch." Three arches, each load-bearing in how an autonomous AI system is governed: an ethical-constraint layer that sets the limits it cannot cross, a moral-direction layer that orients what it should pursue, and a clarity-and-identity layer that keeps it coherent and sure of what it is. Each is necessary; none is sufficient alone. Together they form a structure built to govern complexity without flattening it.