Our in-house engine
Research & Development
Research & Development is Navis Mater's in-house engine. We establish new companies and develop the tools, ways of working and governance that make our portfolio companies stronger.
From research to offer
-
Research
The GAO framework and the flagship white paper: how humans and AI agents can work together under governance you can verify.
OverviewWhite paperGlossary -
Our own operations
Navis Mater runs as the reference implementation. This is where the model is tested in real work.
How we run ourselves ↓ -
The portfolio
What holds up in our own operations is carried into our portfolio companies.
The portfolio → -
The offer
We offer the same way of working to other organizations, as AI agent governance advisory.
See the offer →
How we run ourselves
Our own agent fleet, in operation
- Accountable leadershipSets direction and makes key decisions
- AI Chief of StaffCoordinates the fleet
- Specialist agentsResearch · Engineering · Documentation · Governance work
- Independent reviewReviews the work, outside the coordination; escalates directly to leadership
We run Navis Mater with a fleet of AI agents. They work within their mandates, an AI Chief of Staff coordinates, and consequential decisions go to accountable people. Review sits outside the coordination.
How we govern the fleet
Mandate
Every agent has an explicit mandate: why it exists, what it may do and where it must stop.
Identity & authority
Every agent has its own identity and bounded authority, and cannot widen it on its own.
Controls
Consequential actions are meant to pass enforceable controls, not just policy text. One enforcement boundary is verified; the rest is still being staged.
Evidence
Every consequential action leaves a durable record that can be reviewed afterwards.
Accountability
Named humans remain accountable for consequential decisions. Agents receive authority, not responsibility.
The reference implementation
What we have shown so far
Navis Mater is the reference implementation from which the GAO model is being generalized. The white paper separates what has been demonstrated in operation from what remains to be verified.
- Governed specialist work Real work across research, engineering, review, documentation, coordination and governance. Demonstrated in operation
- Attributable actors Distinct identity per agent, and increasingly scoped, short-lived access. Demonstrated in operation
- Durable coordination Attributable work between agents that survives individual sessions. Demonstrated in operation
- Runtime control evidence One live enforcement boundary, exercised with malicious and benign cases. Demonstrated in operation
- Executive steering Outcomes, priorities, evidence, exceptions and decisions surfaced at management level. Demonstrated in operation
- Early spread beyond the founder Navis Mater’s CTO and controller have started operating their own governed agent teams. Demonstrated in operation
How the white paper was made
- Human direction
- Specialist agents
- Independent review
- Human approval
The flagship white paper was produced through the operating model it describes.
The open question
Does the model keep working without the person who built it?
The paper states the question openly. The first signs: Navis Mater’s CTO and controller now run their own governed agent teams.
R&D FAQ
Questions and answers
What does Navis Mater's R&D do?
Two things: we establish new companies, and we develop the tools, ways of working and governance that strengthen our portfolio companies: CyberResilient, Cyber Defencely, NetSymphony and AceBit.
What is the Governed Agentic Organization (GAO)?
Our flagship: an organization type in which humans and AI agents work together under explicit mandates, attributable identity, bounded authority, enforceable controls, durable evidence and named human accountability.
How is the research used in practice?
Navis Mater is the reference implementation. An AI Chief of Staff coordinates specialist agents working under explicit mandates and bounded authority, independent review checks the work, and consequential decisions are made by accountable people. What holds up in our own operations is carried into the portfolio companies.
What does the AI Chief of Staff do?
The Chief of Staff is an AI agent that coordinates the fleet: it allocates work and keeps priorities together. It does not make the consequential decisions — those stay with accountable people — and review sits outside the coordination.
What have you shown so far, and what remains?
Demonstrated in operation: governed specialist work, a distinct identity per agent, coordination between agents that survives individual sessions, one live control boundary at runtime, steering at management level and early transfer beyond the founder. The open question is whether the model keeps working without the person who built it. The white paper states it openly.