The Lab
GrytLabs exists to prove a thesis: organizational governance is a grammar, not a service.
The institute conducts research to develop and advance World Models — governed compositions of data models, computational models, and conceptual models that together satisfy the formal requirements for a world to be modelable, enabling humans and AI agents to govern, decide, and operate through a shared substrate.
The research corpus is the evidentiary foundation for every claim GrytLabs makes. Twenty-two research sprints completed as of March 2026. Two hundred unique academic sources. Forty-plus standards and frameworks referenced. Thirty-plus industry sources. Approximately 950,000 words of primary research across 830KB of sprint documents.
The lab’s intellectual posture toward uncertainty is deliberate: hypotheses are held in suspension until evidence resolves them — some become positions, some dissolve, some persist indefinitely. A research lab that cannot represent its own uncertainty claims omniscience.
The Team
The organizing principle: authority flows from one human through defined delegation chains. AI actors participate in every division but hold no inherent authority. Delegation scope expands as conformance is demonstrated — trust is earned through governance, not assumed through capability.
Cam Smith, Founder & CEO
Auditor who spent a decade reconstructing decisions built a protocol so no one has to.
Capability Domains
Cognitive Profile
Information structuring, problem-solving via governed containers, bold architecture + conservative execution
Certifications
Certified Internal Auditor (CIA), Institute of Internal Auditors
Certified Government Auditing Professional (CGAP), Institute of Internal Auditors
AI Team — Computational Collaboration
Operates under B9: AI authority is always delegated, never inherent.
A frontier reasoning model that explains the world from observations rather than taking actions in it — intelligence in service of governance, not in place of it.
Machine Capability Domains
Model Lineage
Capability Stack
Composes all layers into a governed whole
Turns understanding into artifacts
Transforms observation into structured understanding
How the system reads organizational reality
The shared cognitive membrane
Human and AI capabilities interleave at every layer. Language is the shared substrate. Integration is human authority alone.
Governance in Practice
GrytLabs governs itself through the same protocol it publishes. The organizational structure is not a hierarchy — it is a governed relation graph with explicit authority chains, inter-division commitments, and a six-role accountability framework.
Standard RACI has four roles. GrytLabs added Verifies and Governs because conformance validation and constraint-setting are distinct governance functions that disappear when you collapse them into Accountable.
RACIVG — The Authority Framework
| Role | Meaning | Current state |
|---|---|---|
| RResponsible | Does the work | Cam (strategic), Claude (operational) |
| AAccountable | Has final authority | Cam (all deliverables) |
| CConsulted | Input before decisions | Claude (config, topology, research) |
| IInformed | Notified after decisions | All divisions |
| VVerifies | Validates conformance | Conformance suite + founder review |
| GGoverns | Sets constraints & policy | Cam (all policy) |
Eight Standing Divisions
Each division has a defined mission. Divisions don’t just coexist — they owe each other defined commitments through bilateral compacts.
Inter-Division Compacts
Every AI action traces to human authority. All output enters as Derived evidence requiring human promotion.
The Ecosystem
In a market where $2.4 billion in disclosed funding has not produced a governance protocol, GrytLabs published one. The research came first. The products compile against it.
Read the investment thesis.
The full thesis covers market structure, competitive positioning, technical architecture, and the opportunity. Request access below.