AVNER — Built on Discipline. Backed by Conviction.

What Avner is

Controlled simulation

Underwriting, capital modelling, reinsurance optimisation, and automated decision systems interact at real-world scale — allowing invisible drift, authority ambiguity, and cascade exposure to be detected, contained, and governed before they destabilise capital.

Structurally real

Every portfolio, committee, capital metric, and AI interaction behaves exactly as it would inside a live specialty carrier — the difference is that it can be observed and stressed safely.

Operational realism

Full staffed down to each member of staff having an email. Realistic scenarios can be run against that activity and demonstrated live.

Company background

Established in 2022 with £750m of cornerstone capital to build a specialty underwriting platform where capital discipline, AI capability, and governance operate as a unified architecture.

Founding thesis

As underwriting becomes increasingly agentic, governance must become constitutional.

Hidden forces inside AVNER

Structural feedback loops

Underwriting AI influenced portfolio mix
Portfolio mix influenced capital model outputs
Capital model outputs influenced reinsurance purchasing
Reinsurance structure altered net exposure concentration
Feedback loops intensified at machine speed
Human oversight operated at meeting speed

Local optimisation, global weakness

Business units adopted AI tools independently
Claims automation optimised for speed
Underwriting models optimised for growth
Capital systems optimised for volatility smoothing
Enterprise coherence weakened globally

None of these forces are failures — they are structural consequences of intelligent automation.

What happens if these forces remain unseen?

Minor bias in cyber risk scoring
Slight underestimation of catastrophe correlation
Gradual increase in delegated underwriting authority
Local optimisation without enterprise visibility

Performance appears stable. Combined ratio remains within tolerance. No alarms triggered.

Portfolio concentration increases subtly in correlated segments
Reinsurance retention adjusted based on model outputs
Capital model absorbs assumptions from drifted underwriting data
Feedback loops tighten across systems

Volatility increases slightly. Capital buffers narrow. Still within declared appetite. Board remains comfortable.

Catastrophe cluster or systemic cyber event
Claims automation escalates loss processing at machine speed
Capital adequacy recalculates under updated assumptions
Retrocession coverage proves misaligned with actual exposure

Now the drift that accumulated silently becomes visible. Combined ratio spikes. Return on equity collapses. Regulators ask questions about AI governance.

Committees meet quarterly
AI systems updated continuously
Post-event forensic analysis reveals decision pathways that were never explicitly authorised

The issue was not malicious AI. The issue was invisible structural drift. The greatest risk of agentic systems is not autonomy. It is unobserved accumulation.

Vision, mission, and pillars

VISION

To become the benchmark for capital-disciplined, AI-governed specialty underwriting across market cycles.

MISSION

To underwrite complex risk profitably and sustainably by integrating disciplined capital allocation, actuarial rigour, and constitutional governance over all human and autonomous decision systems.

Targets

Maintain Solvency Coverage Ratio target ≥ 165%
Preserve cycle-adjusted Combined Ratio < 92%
Align underwriting deployment strictly to declared risk appetite

Strategic & enabling pillars (selection)

Capital Discipline

Solvency II alignment; stress testing under multi-event clustering; capital tolerance governance.

Underwriting Integrity

Independent actuarial validation; continuously test portfolio correlation and tail-risk assumptions.

AI & Agentic Governance

Map AI-assisted decision pathways; real-time authority verification; monitor accumulated uncertainty and cascade risks.

Data & Technology Architecture

Secure AI model lifecycle management; transparent decision provenance; segregation of development and live environments.

Current strategic vision (timeline)

Initial vision (2022)

Align declared risk appetite with actual AI-assisted underwriting behaviour
Detect early-stage model drift in catastrophe and cyber pricing
Map AI-enabled decision pathways across underwriting and claim
Surface accumulated uncertainty across portfolio aggregation

Near-term vision (2026)

Monitor real-time exposure accumulation across correlated risk classes
Implement portfolio-level uncertainty tracking
Prevent cross-system cascade between underwriting, capital modelling, and reinsurance purchasing
Real-time detection of capital misalignment

Future vision (2032+)

Closed-loop capital optimisation with traceable authority
Enterprise-wide epistemic health monitoring
Regulatory-grade transparency across all automated decisions
Embed enforceable authority boundaries for autonomous underwriting agents

Office locations

Global HQ — London

Board Executive Committee
Underwriting Leadership
Finance & Treasury
Risk & Exposure Oversight
Compliance & Regulatory Affairs
Broker Relationship Management

Capital & Reinsurance Hub — Bermuda

Sidecar Structures
Retrocession & Reinsurance Purchasing
Capital Management
Third-Party Capital Relations
Solvency & Stress Modelling

Technology & AI Governance Lab — Winchester

Data Engineering
AI Model Lifecycle Management
Decision Provenance & Audit
Underwriting Automation Oversight
Cyber Security Operations

North American Specialty — New York

US Specialty Underwriting
Broker Distribution
Regulatory Interface
Claims Oversight

Modelling & Analytics Centre — Zurich

Catastrophe Modelling
Cyber Risk Analytics
Capital Model Development
Exposure Aggregation
AI & Model Governance

Indicative realistic scenarios

Helios

Declared Risk Appetite: Peak US wind exposure capped at 18% capital
Observed Exposure: 22% due to AI-assisted underwriting bias
Sentinel flags authority boundary breach
Atlas shows cascade effect into: reinsurance cost, capital buffer reduction, portfolio volatility

Hydra Cyber

AI risk scoring updated silently
Accumulated uncertainty increased 1.8% loss ratio drift
Capital model absorbed assumption
Atlas identifies correlation blind spot
Sentinel flags model version drift
Board sees potential combined ratio impact

This simulates governance failure containment (demo).

Current status of Avner

Helios combined ratio: 91.8% (first full underwriting year); peak US wind aggregation increased 3.4% above declared appetite due to accelerated AI-assisted risk selection.
Hydra GWP: +28% YoY; independent actuarial review indicates early signs of model variance emerging in long-tail assumptions.
Capital coverage ratio stable at 167%; internal stress testing shows volatility sensitivity increasing under multi-event clustering scenarios.
Reinsurance optimisation under Boreas reduced net retention volatility by 6%; cross-portfolio correlation mapping remains partially manual and quarterly.
AI-assisted underwriting active across Property, Cyber and Specialty; authority boundaries are defined, but exception escalation and cross-system cascade visibility remain limited.
Exposure aggregation platform upgrade (Sirius) is 80% complete; full real-time portfolio concentration monitoring expected within 9 months.
Independent Model Validation review identified incremental drift in catastrophe recalibration assumptions following recent climate-adjusted updates.
Board requested enhanced visibility into automated decision pathways, authority exceptions, and capital-at-risk under AI-driven optimisation scenarios.

Avner is operationally strong and capital-resilient; however, increasing decision velocity and distributed AI deployment have created emerging governance blind spots requiring real-time oversight and authority containment.