Insurance Examination

Enforcing Agency as a Risk Dimension

The NAIC's Supplement defines agentic AI by what a system can do rather than how well it predicts. Dorvelon's filed comment sets out the dimensions — data access, action authority — a risk classification must declare.

Governance Briefings

Standards in Formation

Proving Documentation Holds as Evidence

NIST is establishing the first public standard for AI documentation. Dorvelon's submitted position: in regulated deployment that documentation is consumed as evidence, and evidence carries three properties beyond content.

August 2026

Agentic AI Assurance

Moving Assurance from Document to Live Control

Agentic AI now acts on live infrastructure faster than governance can review, and documentation no longer proves control. Why high-materiality agentic AI requires runtime assurance, not documentary assurance alone.

June 2026

Machine-Speed Governance

Securing the Institutional Mandate Over Agentic AI

An executive briefing on why paper-speed governance fails against agentic AI, drawn from the P.A.S.S. risk taxonomy, three real-world incident diagnoses, and the seven-layer audit-ready blueprint.

June 2026