Making Agentic AI insurable.

Specialist governance and risk advisory for carriers, syndicates and reinsurers.

Seattle

01 — Brief

The risk surface is shifting.

AI agents are running enterprise functions: negotiating contracts, settling claims, managing procurement, handling customer service, and deploying production code. Anthropic's 2026 State of AI Agents Report finds 57% of enterprises running agents in multi-stage workflows, including 16% spanning multiple functions.

For carriers, this is a new risk category. The underwriting frameworks built for human-operated businesses do not yet have the vocabulary, the controls, or the actuarial basis to price it. We help carriers build them.

Agent decision failure

ii.

Model hallucination, fabrication & deception

iii.

Data protection & privacy breach

iv.

Algorithmic discrimination

Cyber & prompt-injection exposure

vi.

Regulatory violation & drift

vii.

Faulty delegation of authority

viii.

Multi-agent cascade failure

ix.

Unauthorized tool access and execution

02 — Practice

Five lines of counsel.

Agentic AI risk assessment

Structured evaluation across seven dimensions: technical, governance, data protection, cyber, regulatory, liability, and accumulation. Output is a risk profile for the underwriting committee.

ii.

Governance frameworks

Frameworks for carrier and insured: roles and responsibilities, human-in-the-loop design, incident response, vendor governance, and model lifecycle.

iii.

Underwriting enablement

Direct support to underwriting teams: defining insurable AI risks, drafting exclusions, structuring policies, classifying risk, and setting minimum control requirements for insureds.

iv.

Regulatory & legal mapping

Mapping across EU, UK and US AI regulations, privacy regulations, cybersecurity statutes, and operational resilience frameworks.

Governance certification

Certification models, governance maturity benchmarks, and a market standard for insurable AI governance — alongside an AI risk rating platform for syndicates.

— Forthcoming

03 — Method

How an engagement runs.

Architecture review

System architecture, degree of autonomy, data access structures, decision-making authority — documented before any control assessment.

Governance blueprint

Governance handbook, role and responsibility model, control framework, incident protocols — built around the carrier's risk appetite.

iii

Underwriting support

Risk scoring, AI risk taxonomy applied to the book, insurability criteria, and minimum control requirements.

Incident simulation

Closed-loop simulation of AI loss scenarios: erroneous contract execution, data breach, incorrect claims decision, agent operational failure.

04 — Counsel

Founder-led.

Two founders — one in law, one in AI. They run the engagements.

Co-Founder — Senior Legal Counsel

Dr. Matthias Artzt

Financial services, AI & data protection

Extensive professional experience as an attorney and in-house counsel in the financial services industry, with a focus on artificial intelligence and data protection. Speaker, editor and author of numerous specialist publications in these fields. Work spans governance frameworks, policy structures and risk management.

Co-Founder — AI Pioneer

John deVadoss

Frontier AI, machine learning & cloud computing

Frontier AI researcher and start-up founder (successful enterprise AI exit), having multi-decade enterprise software engineering background, with a focus on cloud computing. Author of multiple books on software architecture, frequent speaker at global technology forums. Work spans machine learning, agentic architecture, and recursive AI.

05 — Publications