Insurance Law · Governance · Technology

How better judgment is built — and where it breaks down — in the face of complexity, and how well-designed governance enables success

I am a senior in-house lawyer in the London specialty insurance market with over fifteen years’ post-qualification experience, including seven years at Lloyd’s syndicates and specialty carriers.

My professional focus is the intersection of insurance law, governance, and emerging technology. I build legal infrastructure — wordings portfolios, technology programmes, governance frameworks — and I study how institutional decision-making actually operates under the pressures of commercial constraint, regulatory expectation, and incomplete information.

This site hosts two expressions of that work: interactive governance scenarios that reconstruct realistic decision environments for senior insurance professionals, and Legal Debugged, a blog exploring AI, legal technology, the law and institutional judgment.

Recent Work

Legal Debugged · New

Who’s in the Room When You Talk to AI?

Privilege after United States v. Heppner — the first US ruling on attorney-client privilege and direct AI interactions. What it means for enterprise legal teams, AI governance, and the future of confidentiality.

Read the analysis

Governance Scenario

Model Behaviour

A pricing algorithm for emerging market infrastructure risks performs well in backtesting. Limitations are documented. Caveats are clear — initially. An interactive exploration of how institutional governance operates when technology displaces established expertise.

Explore the scenario

15+

Years PQE

7

Years In-House

Lloyd’s

Specialty Market

Cert CII

Insurance Qualified

WCI

Freeman

About

Most governance failures in insurance do not arise from ignorance, bad faith, or technical error. They arise from how judgment is exercised inside institutions: how decisions are framed, how caveats erode as they move through governance structures, how informal signals acquire authority, and how responsibility becomes distributed just enough that no individual feels accountable — until outcomes demand explanation.

My work examines those patterns, drawing on over fifteen years’ experience across contentious, commercial, and advisory environments in the London specialty insurance market — a market defined by complexity, asymmetry of information, and sustained regulatory scrutiny, where hindsight is unforgiving and accountability is rarely binary.

I am a solicitor with seven years in-house at Lloyd’s syndicates and specialty insurers, advising board-level stakeholders across wordings strategy, technology, AI, cyber risk, and regulatory responses spanning Financial Lines, Casualty, Cyber, and Kidnap & Ransom. Before moving in-house, I spent seven years in private practice as an insurance and reinsurance litigator, managing multi-million pound disputes across professional indemnity, coverage, and cross-border reinsurance claims.

I carry a conviction that legal functions should be built like the best products: data-driven, scalable, and enablers of commercial growth — not gatekeepers to it. That conviction has shaped how I approach legal technology, from securing investment and building business cases for AI adoption to designing workflows that give legal teams the data they need to advise strategically rather than reactively.

My work sits at the intersection of insurance law, governance, and emerging technology — and it is that intersection which the governance scenarios and writing on this site are designed to explore.

The governance scenarios are structured reconstructions of realistic decision environments. They are not case studies, training exercises, or assessments. They are designed for senior professionals who recognise that formal frameworks are necessary but insufficient, and that many of the most consequential governance decisions occur in the space between policy, process, and practice.

Legal Debugged, the blog, explores AI, legal technology, the law and institutional judgment. The focus is practical: how emerging legal and regulatory developments affect the way in-house teams operate, advise, and build.

The premise across both is simple: it is better to examine how judgment is exercised deliberately and honestly now than to have it examined later through commercial failure, investigation, or regulatory intervention.

Qualified Solicitor (England & Wales), admitted 2011. Business-level French.

Governance Scenarios

These scenarios reconstruct realistic decision environments drawn from years of in-house practice in the London specialty insurance market. They explore how institutional risk emerges through ordinary decision-making — how language migrates through committees, how individually reasonable choices aggregate into governance problems, and how documentation creates one institutional reality while lived experience creates another.

They are designed for senior audiences: board members, underwriting leadership, legal, risk, compliance, and governance professionals who operate in environments where information is incomplete, incentives are misaligned, and accountability is distributed.

The patterns explored here — model risk governance, conduct compliance, the gap between documentation and institutional reality — are the patterns that a well-designed legal function exists to anticipate and address. Understanding them is a precondition for building one.

There are no right answers. Only the paths taken — and what they reveal.

Interactive scenario

Model Behaviour

Algorithmic underwriting and the governance of institutional judgment

Learn more

Interactive scenario

Commission by Omission

Fair value and the distance between documentation and institutional reality

Learn more

Interactive scenario

Model Behaviour

Algorithmic underwriting and the governance of institutional judgment

Interactive narrative · ~25 minutes · Five decision points

Overview

You built a pricing algorithm for emerging market infrastructure risks. The backtesting is promising — a Gini coefficient of 0.42 on a holdout sample of 180 risks. Enough to demonstrate discriminatory power. Not enough to give confidence in tail behaviour.

Your validation report documents the limitations. The version circulated to the Underwriting Committee leads with the positives. Technical details sit in Appendix C. Committees rarely read appendices.

Over thirty months, the model succeeds in Latin America, is extended to West Africa against your documented concerns, and displaces a senior underwriter whose instincts were correct but whose interventions made things worse. You are promoted. He leaves. The institution learns. The cost is real.

Context

  • Sample size: 180 risks
  • Validation report: edited before committee circulation
  • Governance framework: PRA SS1/23 requires documentation of limitations

The model works. The question is what “working” means — and what it costs.

Themes

Model risk governance Technical communication under pressure Documentation versus reality Expertise displacement Institutional learning
This scenario is designed for reflection, not assessment. There are no right answers — only the ones you choose, and what they reveal about how you navigate the space between technical rigour and institutional reality.

Interactive scenario · ~25 minutes

Launch Model Behaviour

Opens in a new page. Five decision points. No data collected.

Beta

This scenario is currently in beta. Feedback is welcome.

Contact gregorybutera@tuta.io for early access or collaboration opportunities.

For analysis of AI, privilege, and legal technology in insurance, see Legal Debugged.

Interactive scenario

Commission by Omission

Fair value and the distance between documentation and institutional reality

Interactive narrative · ~20 minutes · Multiple reflection points

Overview

Your infrastructure PI book has been profitable for three years. Combined ratio in the low sixties. The sort of performance that earns quiet approval and keeps questions to a minimum.

When the market turns, broker conversations shift. Threshold accommodations accumulate — each documented, each within authority, each individually unremarkable. By November, Henderson’s effective commission rate is running at 38%. You do not know this yet.

The Conduct team’s fair value assessment asks a question you would rather answer verbally. A governance review is commissioned. No individual transaction breaches policy. The aggregate position is different from the sum of its parts. Your promotion is deferred. Nobody tells you that you did anything wrong.

Context

  • Contingent commission: common in specialty lines, requires fair value assessment
  • Threshold accommodations: individually reasonable, collectively material
  • File documentation: accurate but incomplete
  • Governance review: commissioned six months after pattern emerges

Themes

Fair value compliance Incremental decision-making Language and institutional record Informal pressures and formal authority What gets documented versus what gets remembered
A reconstruction of how individually reasonable paths lead to collective governance questions. The documentation shows one story. Your memory holds another. The review will find a third.

Interactive simulation loading area

The Commission by Omission simulation will load here.

Beta

This scenario is currently in beta. Feedback is welcome.

Contact gregorybutera@tuta.io for early access or collaboration opportunities.

For analysis of AI, privilege, and legal technology in insurance, see Legal Debugged.

Legal Debugged

Analysis at the intersection of law, technology, and institutional decision-making in specialty insurance. Written from the perspective of in-house practice in the London market.

Who’s in the Room When You Talk to AI? Privilege After United States v. Heppner

A New York Federal Court has ruled that a criminal defendant’s conversations with Claude, Anthropic’s AI assistant, are protected by neither attorney-client privilege nor the work product doctrine. The ruling appears to be the first in the US to address privilege claims over direct AI chatbot interactions. What it means for enterprise legal teams, AI governance, and the future of confidentiality — and why English lawyers should be paying attention.

For an interactive exploration of how institutional governance operates under pressure in specialty insurance, see the Governance Scenarios.

Get in Touch

I welcome conversations about governance, legal technology, and decision-making in complex insurance environments — whether that is a specific engagement, a speaking opportunity, or an idea worth exploring.

Location

London, United Kingdom

Areas of Interest

  • Design and build of legal functions for complex, multi-jurisdictional insurance environments
  • Legal technology strategy and AI governance for in-house teams
  • Wordings portfolio management and governance
  • Legal and regulatory risk in specialty and emerging markets
  • Governance design for institutional decision-making