Using GenAI to increase trust and transparency in motor claims

Crossroads ahead

The motor insurance industry in the GCC is approaching an inflection point. For decades, the fundamental economics of claims — the relationships that govern repair decisions, the opaqueness that surrounds pricing and quality, the manual processes that determine outcomes — have been sustained by a simple reality: there was no viable alternative.

The tools to capture, structure, and act on claims data at scale did not exist. The cost of the current model, while significant, was invisible because nobody had the means to measure it.

That is no longer true.

Artificial intelligence — and specifically the combination of large language models, computer vision, and structured data architecture — now makes it possible to do what the industry has talked about for a decade but never operationalized: build claims processes where every decision is evidenced, every cost is benchmarked, every communication is logged, and every stakeholder, from the customer to the regulator, can see exactly what happened and why.

This is not a technology dissertation. It is a paper about trust, and about the business case for making trust measurable. The insurers that move first will not just reduce fraud and leakage; they will build a structural advantage in customer retention, regulatory standing, and operational resilience that compounds over time. Insurers who wait will continue absorbing costs they cannot see, defending decisions they cannot evidence, and losing customers they never understood.

This paper lays out the case in eight chapters: why trust is breaking down, what genuine transparency looks like, how to build an AI trust stack with today's imperfect data, how governance keeps humans in control, where the highest-impact use cases are, what to measure, how to get started, and what to watch out for along the way.

‍

Trust Is breaking down

Opaque decisions, inconsistent handling, slow outcomes, and rising fraud

Trust in GCC motor claims is breaking down because the economics and expectations of the market have modernized, while the repair ecosystem and many claims practices have not.

For decades, repair decisions have been mediated through an "old-guard" network: seasoned claims handlers relying on long-standing personal relationships with preferred garages and informal price norms. That model can function when volumes are low, scrutiny is limited, and decision-making is largely trusted by default. It fails in today's environment, where claim severity is rising, vehicle complexity is increasing, fraud is more organized, and regulators and customers expect provable fairness.

Even today, repair scopes and costs still originate as handwritten estimates, negotiated behind closed doors, and benchmarked against the same tight circle of workshops. This is not only inefficiency; it is an evidence gap. Insurers struggle to demonstrate that prices are market-consistent, that repair quality is verified, that decisions are free from bias or conflicts, and that leakage is actively controlled rather than retrospectively explained.

‍

Key drivers eroding trust in motor claims (GCC context)

• Evidence gap in pricing and quality: Paper-based estimates, limited photo/parts traceability, weak benchmark data, and inconsistent standards of repair validation.
• Customer expectations shifting fast: Customers now expect real-time visibility, predictable timelines, and clear explanations — not phone calls and ambiguity.
• Network effects that entrench opacity: "Comparable quotes" sourced from the same ecosystem, informal referral patterns, and limited competitive tension.
• Regulatory and conduct pressure: Higher expectations for documented decision rationale, complaint handling, service standards, and demonstrable control frameworks.
• Rising severity and complexity: Advanced driver-assistance systems, sensors, calibration needs, and higher parts costs increase variance and dispute risk.
• Talent and scalability constraints: Scarcity of modern claims analytics capability; heavy reliance on a few experienced individuals creates key-person risk and inconsistent outcomes.
• Fraud and organized leakage: Inflation of labor hours, parts substitution, add-ons, duplicate invoicing, staged damage, and referral economics that are hard to detect without data.
• Fragmented supply chain with limited traceability: Multiple intermediaries (recovery services, car rentals, parts suppliers, sub-contracted specialists) each add cost and complexity with little end-to-end visibility or accountability.

This is a clear and present balance-sheet, conduct, and brand risk problem compounding over time.

When decisions can't be evidenced end-to-end, leakage becomes structurally embedded in the loss ratio, disputes and reopenings rise, and cycle times stretch. The old model doesn't scale: it depends on individual judgment and relationships rather than repeatable processes and verifiable data. In a market where customers expect transparency and regulators expect proof, insurers either industrialize trust through data, controls, and auditability — or they accept a persistent "trust discount" in profitability, compliance posture, and reputation.

‍

Board-level Implications

• Limited profitability: Severity inflation and hidden leakage become "normal," reducing the ceiling on combined ratio improvement.
• Conduct risk: Inconsistent outcomes and weak decision traceability increase audit findings and complaint escalation.
• Operational resilience: Key-person dependency and informal practices create fragility and inconsistent quality.
• Customer experience: Low visibility and slow resolution erode retention and raise acquisition costs.
• Strategic positioning: Insurers fall behind peers who can prove fairness, speed, and control with evidence.

‍

What "transparency" mneans in insurance

Traceability, explainability, consistency, and customer visibility

Before exploring how AI can help, it is worth defining what genuine transparency looks like in motor claims — not as it exists today, but as the standard the industry must be working towards. In motor claims, transparency has three concrete dimensions: insights, explainability, and consistency.

‍

Giving the customer insight into the repair

The biggest source of customer frustration is the lack of information. Once a vehicle enters the workshop, the policyholder typically enters an information vacuum: no clarity on the repair, what parts are being fitted, what quality checks are performed, or when the car will be returned.

Every repair already produces data: parts orders, labor records, inspection notes. The problem is that none of it reaches the customer. It sits in disconnected systems, in different formats, often in different languages.

Generative AI can ingest this data from workshops, parts suppliers, and quality checkpoints, and translate it into plain-language updates pushed directly to the policyholder: which garage has their car, what parts are being used, what stage the repair has reached, and what controls have been completed. And equally important, the same information creates an auditable record for the insurer — what serves the customer also serves the claims file and the compliance function.

Customer-facing transparency includes:

• Plain-language repair status updates pushed to the policyholder at each stage
• Visibility on the workshop handling the repair, including performance history
• Itemized parts information: OEM vs. aftermarket, supplier, and fitment confirmation
• Quality control evidence: photos, checklists, and sign-off records shared with the customer
• Every communication logged as part of the claims audit trail

‍

Explaining decisions, not just communicating them

Transparency also means explaining outcomes, particularly when the insurer limits what it will cover. A common dispute arises when pre-existing damage is identified during repair. Traditionally, a customer is told that certain damage "is not related to this claim" with little supporting evidence. From the customer's perspective, this feels arbitrary and frustrating.

AI-powered inspection tools offer a path forward. When a vehicle is assessed at the point of claim using structured photo capture and computer vision, its condition can be documented comprehensively. If the insurer can later show the customer a clear, time-stamped visual record that specific damage existed before the incident, the conversation shifts from confrontation to explanation.

Explainability includes:

• AI-assisted condition assessments at FNOL, creating a verifiable baseline
• Visual evidence (time-stamped, geo-tagged) distinguishing claim damage from pre-existing wear
• Plain-language explanations linking each coverage decision to specific evidence
• A single evidence base serving the customer, the claims file, and any future dispute resolution

‍

Consistency: from individual transparency to systemic trust

Customer visibility and explainability address individual claims. Transparency becomes truly powerful when it is systematic — when every claim follows the same verifiable path.

When every FNOL follows a standardized intake, every inspection uses the same AI-guided protocol, every repair is tracked against the same milestones, and every quality check is logged against the same criteria, the process becomes auditable by design. This is the "sunlight effect": when every participant knows that every action is recorded and auditable, behavior self-corrects. Fraud prevention becomes embedded in process architecture, not bolted on as retrospective detection.

Systemic consistency includes:

• Standardized processes across every claim, removing variance caused by individual judgment or relationships
• A complete audit trail demonstrable to regulators, reinsurers, and board governance at any time
• Workshop accountability through continuous, data-driven performance monitoring
• Natural fraud deterrence through tracking and tracing of every decision and transaction
• Operational resilience: the process no longer depends on key individuals or informal knowledge

‍

The AI trust stack: getting started with what you have

Data integrity, decision guardrails, auditability, and customer communications

Conversations about AI and data in insurance stall because the gap between the current state and the ideal feels insurmountable. Data is fragmented, inconsistent, mostly unstructured. Workshop systems are basic or non-existent. Internal processes vary by team, by handler, by day of the week. And so the conclusion is: "we're not ready."

That conclusion is wrong. The AI trust stack is not an all-or-nothing investment. It is a set of layers that can be built incrementally, each one adding value on its own while creating the foundation for the next.

‍

The four layers

• Layer 1 — Data integrity: Minimum data model with quality gates — not perfect data, but consistent, structured, and verifiable. This is the foundation: making sure the right data is captured, in a consistent format, at the right points in the claims journey. Large language models can extract meaningful, structured information from messy, imperfect sources — handwritten estimates, unstructured emails, voice notes, unlabelled photos. Generative AI handles extraction; deterministic programming handles validation and structuring.
• Layer 2 — Decision intelligence: AI-powered triage, benchmarking, and anomaly detection, effective even on imperfect data. Once consistent data is flowing in, AI does what it does best: pattern recognition, anomaly detection, and triage. Are the labor hours on this estimate in line with comparable repairs? Does the parts list match the documented damage? Has this workshop shown a pattern of inflated scopes? AI is remarkably effective at surfacing inconsistencies and outliers even in imperfect data.
• Layer 3 — Auditability: Automatic, end-to-end logging of every data point, flag, and decision as a byproduct of digital process. When data capture and decision support are digitized, auditability comes almost for free. This is the audit trail that regulators expect, that reinsurers value, and that protects the insurer in disputes — not a separate system to build, but a natural byproduct of Layers 1 and 2.
• Layer 4 — Customer communications: Generative AI translating repair data into clear, proactive policyholder updates. With structured data and an auditable process, generating clear, accurate customer updates becomes straightforward. The insurer can explain what is happening, why, and what comes next — because the underlying evidence exists.

‍

Starting small, but starting now

The minimum viable version of this stack is not a multi-year transformation program. It starts with defining the data you need for each claim, and insisting on getting it: a structured FNOL intake with mandatory photo capture, a standardized estimate format, digital confirmation of parts used and work completed, basic milestone tracking from assignment to delivery.

The enrichment comes over time. As data accumulates, benchmarks emerge. As benchmarks emerge, anomalies surface. As anomalies are investigated, processes tighten. The insurer that starts capturing structured data today will, within months, have a dataset that enables meaningful AI-driven insights — even if the starting point was a blank page.

‍

The mninimum data model

• At FNOL: Structured photo set (minimum angles defined), standardized damage description, vehicle identification and condition baseline
• At estimate: Digital estimate in a comparable format, itemized parts and labor, benchmark-ready pricing
• During repair: Milestone updates (parts ordered, repair started, quality check, completion), parts confirmation (OEM/aftermarket, supplier)
• At delivery: Completion photos, quality inspection sign-off, customer confirmation
• Throughout: Every data point time-stamped, geo-tagged where relevant, and logged to the claims audit trail

‍

Governance and compliance-by-design

Humans stay in control. The system proves it.

A reasonable concern with any AI-driven process is: who is actually making the decisions? The answer, in a properly designed trust stack, is straightforward: humans do. AI surfaces information, flags anomalies, and accelerates workflows. But at every critical juncture, deterministic logic governs what happens next — and that logic is designed to keep humans in the loop where it matters.

In practice, this means threshold-based approval gates built into the claims workflow. If an estimate exceeds a defined value, the process pauses and routes to a senior handler for review. If an AI flag identifies a potential fraud indicator, the claim is escalated to an investigator, not auto-declined. If parts costs deviate from benchmark ranges, a human reviews before authorization. These are not AI decisions; they are rules-based checkpoints, coded in traditional programming, that determine when the process continues automatically and when it stops for human judgment.

What compliance-by-design delivers:

• Threshold-based approval gates that pause the process for human review at defined trigger points
• Every decision, data point, and AI flag logged with timestamps and user attribution
• Full claim traceability recoverable in seconds, not days, for any regulator, auditor, or dispute
• Segregation of duties enforced by system logic, not by policy documents alone
• A defensible record that the insurer followed its own processes, consistently, for every claim

For financial services institutions operating under increasing regulatory scrutiny, this is not a nice-to-have. It is the difference between being able to demonstrate control and hoping that control existed.

‍

High-impact use cases: the trust stack in action

Claims, underwriting, and customer service

Claims: evidence-first orchestration

This is the primary use case. The objective is to replace relationship-based handling with an evidence pipeline that produces three things the old model cannot: a consistent baseline of vehicle condition so disputes don't become opinion battles, comparable repair scope and pricing across workshops so benchmarking is real rather than social, and a machine-readable claim record where every change to scope, parts, labor, and approvals is traceable.

At FNOL, computer vision classifies and tags damage images while LLMs extract structured data from whatever format the workshop submits. During assessment, AI benchmarks the estimate against comparable repairs and flags anomalies: labor hours above peer norms, parts lists that don't match documented damage, pricing inconsistencies, patterns associated with a specific workshop. Not decisions — but flags for review, routed to the right person at the right time through the approval gates.

What evidence-first claims orchestration delivers:

• Structured extraction from any input format (photos, PDFs, handwritten estimates, emails) into a comparable data model
• Real-time anomaly detection: scope inflation, parts substitution, labor hour outliers, repeat patterns by workshop
• Fraud signals assembled as an evidence pack with explainable pointers, not accusations
• Proactive customer updates derived from the same auditable record the insurer relies on
• Every change to scope, parts, approvals, and settlement logged and traceable

‍

Underwriting: explainable risk and pricing governance

The underwriting version of opaqueness is inconsistent decisions, undocumented judgment, and weak traceability of why a quote moved. The same trust stack principles apply, adapted to a different workflow.

AI acts as the document intelligence engine. Underwriting submissions arrive as PDFs, spreadsheets, emails, and broker notes. LLMs extract insured details, vehicle specifications, usage, claims history, and coverage requested into a structured submission record. At the decision intelligence layer, AI surfaces risk signals and governance triggers, spots outliers in frequency and severity patterns, and flags incomplete disclosures. For pricing, it suggests placement within corridors based on comparable risk profiles and forces structured rationale capture when a human deviates from norms. AI cannot bind or alter coverage; it can only propose and explain.

What explainable underwriting delivers:

• Structured extraction from submissions, reducing manual re-keying and catching conflicts across documents
• Risk signals and anomaly detection surfaced before the underwriter commits
• Pricing governance: deviations from corridors are visible, justified, and logged
• Broker and customer-ready explanations of what drives terms and what actions could improve them
• Full quote-to-bind traceability for conduct reviews and audits

‍

Customer service: proactive transparency

The escalation cycle that erodes trust is familiar: slow updates lead to frustration, frustration leads to complaints, complaints lead to disputes and reopened claims. This use case breaks that cycle by making transparency an always-on service rather than a reactive exercise.

The customer-facing assistant reads from the same structured claim record created in the claims use case. It does not invent; it retrieves and summarizes. When a customer asks "where is my car," the assistant answers with facts: the workshop handling the repair, the current stage, the parts ordered, the estimated completion. AI monitors for complaint risk: sentiment patterns across interactions, SLA breaches, stuck milestones, high-friction moments like scope reductions or parts delays.

What proactive customer service delivers:

• Accurate, evidence-grounded answers to customer queries drawn from the structured claim record
• Proactive updates pushed at key milestones without the customer needing to chase
• Complaint risk detection and early intervention before frustration escalates
• Consistent quality of explanation regardless of channel, time, or handler availability
• Every interaction logged as part of the compliance and audit framework

These three use cases are not independent projects. They share the same data layer, the same governance framework, and the same audit architecture. An insurer that builds the trust stack for claims automatically creates the foundation for explainable underwriting and proactive customer service.

‍

Implementation roadmap

Start small. Start now. Scale on evidence.

The reason nothing changes is not lack of awareness; it is inertia. The current model works in the narrow sense that claims get paid, workshops get used, and customers mostly don't leave. The pain is diffuse — leakage spread across thousands of claims, fraud that is never detected, quality failures that surface as complaints months later — rather than acute. The cost of the current model is invisible precisely because nobody is measuring it.

Phase 1: See what you've been missing (Days 1–90)

This is not a transformation program. It is a proof of concept on a contained subset of claims — a single line of business, a specific region, or a defined volume. Implement structured FNOL data capture with mandatory photo sets. Apply basic AI extraction to convert whatever the workshops submit into structured data. Run the minimum data model. The only question you are answering is: what does the data reveal that you couldn't see before?

The answer will be significant. Patterns in estimate inflation, inconsistencies between damage photos and repair scopes, pricing variance across workshops, missing milestones — all of it becomes visible for the first time. Low cost. Low risk. High signal.

Phase 2: Benchmark and measure (Months 4–9)

With structured data flowing, turn on the decision intelligence layer. AI begins benchmarking estimates, flagging anomalies, and triaging claims by complexity and risk. Start tracking the KPIs that matter most: FNOL evidence completeness, flag precision, estimate scope accuracy, cycle time by stage. This is also the phase to begin the workshop conversation in earnest. The first financial signals appear here: leakage identified and quantified, scope anomalies flagged before approval, cycle time patterns that explain cost overruns.

Phase 3: Operationalize (Months 9–18)

Scale the trust stack across the participating portfolio. Customer communications layer goes live: proactive updates, evidence-based explanations, milestone tracking visible to the policyholder. Governance framework fully embedded: approval gates, override logging, compliance-by-design audit trail. Scale decisions beyond this point are driven by evidence from Phases 1 and 2, not by projections or assumptions.

‍

The harsh reality

Most GCC insurers do not have the internal technology team, workshop management capability, or process design expertise to build this themselves. And they should not need to. The realistic path for most is a TPA partner that brings the trust stack ready-made: the insurer provides the portfolio, the partner provides the infrastructure, the data discipline, and the workshop accountability. The insurer retains oversight, governance, and strategic control. The partner delivers the execution capability.

Every month without structured data is another month of invisible leakage, undetected fraud, and unreported quality failures. That cost does not sit in a single line item; it is embedded in every loss ratio that cannot be fully explained, every complaint that could have been prevented, and every regulatory review that depends on reconstructing decisions after the fact.

‍

KPIs for trust and transparency

What to measure, and why it matters

Transparency is only meaningful if it can be measured. The following KPIs give insurers a practical scorecard for tracking whether the trust stack is delivering results across the full claims lifecycle. None require perfect data to start tracking; most can be baselined from existing operations and improved as the data foundation matures.

Data integrity

• FNOL evidence completeness rate: % of claims with the minimum required photo set and mandatory fields captured at first contact. Predicts dispute rate, rework, and cycle time downstream.
• First-time-right data rate: % of claims requiring no follow-up for missing or incorrect core data (VIN, driver, incident details, photos). Direct driver of cycle time and cost-to-serve.
• Workshop data compliance score: Quality-weighted compliance with the required data model: structured photos, estimate format, milestone updates, parts confirmation.

Decision intelligence

• Flag precision and yield: % of AI flags that lead to confirmed leakage, fraud, or adjustment. Yield measured as value recovered or avoided per 1,000 claims.
• Estimate scope accuracy: Gap between initial estimate and final settled scope (excluding genuine hidden damage supplements).
• Supplement rate and severity: Frequency and size of supplements after repair starts. High rates signal weak early evidence capture or strategic under-scoping by workshops.
• Leakage avoided / recovered: Monetary value of scope reductions, pricing corrections, duplicate detection, and recoveries triggered by AI-surfaced anomalies. The core financial ROI measure.

Auditability and governance

• Decision traceability score: % of key decisions with a complete rationale chain: evidence → rule or policy basis → approver. The strongest governance KPI in this framework.
• Override rate (and justified override rate): How often handlers override AI flags or benchmark guidance, and whether they document why.
• Approval gate compliance: % of claims exceeding thresholds that properly trigger review and approval.
• Audit retrieval time: How quickly a full decision trail can be produced for any given claim. In a compliance-by-design environment, this should be seconds, not days.
• Regulatory and audit findings: Number and severity of findings in internal audits, regulatory reviews, and reinsurer assessments.

Customer transparency and experience

• Proactive update coverage: % of claims where the customer received milestone updates without asking.
• Inbound status-chasing contact rate: Customer contacts per 1,000 claims that are pure "where is my car?" inquiries. If transparency is working, this drops sharply.
• Time-to-first meaningful update: Hours from FNOL to first customer communication containing concrete next steps.
• Dispute rate and resolution time: How often customers dispute scope or coverage decisions, and how fast those disputes are resolved.
• Complaint rate (per 1,000 claims): The broadest measure of customer trust. A declining rate indicates that transparency, explainability, and proactive communication are working together.
• Customer NPS / satisfaction score: A lagging but important indicator. Improvements should translate into measurable shifts in sentiment over time.

Repair quality and operational performance

• End-to-end repair cycle time (FNOL to vehicle delivery): The single most visible operational measure for both insurers and customers. Decomposed by stage to identify where AI and process discipline are moving the needle.
• Comeback rate (repair rework): % of repairs returning for defects within 30/60/90 days.
• Quality checkpoint pass rate: % of repairs passing quality control on first inspection.
• Parts lead-time variance: Variation between expected and actual parts arrival. A major driver of customer frustration and rental cost overruns.
• Rental / loss-of-use days per claim: Hard financial impact directly tied to repair delays and process efficiency.

Financial and portfolio (board-level)

• Leakage ratio: Estimated leakage as % of paid losses (scope inflation, pricing variance, parts substitution). Tracks structural improvement, not just individual wins.
• Cost-to-settle per claim: Operational cost (handling touches, vendor interactions, rework) per claim. Structured workflows and AI triage should reduce this progressively.
• Claims severity variance: Variance around expected severity for comparable claim types. Tightening variance indicates consistency and control across the portfolio.

‍

Common failure modes and guardrails

What can go wrong, and how to prevent it

AI in insurance fails predictably when governance is weak. These are the failure modes that appear most often, paired with the guardrails that prevent them.

Hallucinated explanations

Generative AI can produce outputs that read convincingly but are factually wrong — fabricating rationale, inventing data points, or citing evidence that doesn't exist. In a claims context, this is dangerous: a hallucinated explanation sent to a customer or used in a dispute becomes a liability, not an asset. The mitigation is architectural: customer-facing and decision-support AI must operate in "grounded generation" mode, where every output must reference and be derived from structured claim data. Free composition should never be permitted for explanations, coverage decisions, or audit-facing content.

Alert fatigue

When AI generates too many flags, most of which turn out to be noise, handlers learn to ignore them. This is worse than having no flags at all, because it creates the illusion of oversight while real anomalies pass through unchallenged. The mitigation is continuous measurement and tuning: flag precision must be tracked as a KPI, thresholds adjusted regularly based on outcomes, and low-precision signals retired or redesigned rather than left running.

Automation bias

The opposite of alert fatigue: handlers trust AI output too readily and stop applying their own judgment. Over time, the human-in-the-loop becomes a rubber stamp. The mitigation is structural: threshold-based approval gates force genuine review at defined trigger points, and override decisions must be logged with documented rationale. Periodic audit sampling should specifically test whether handlers are engaging critically with AI outputs or simply confirming them.

Privacy leakage

In a system where multiple stakeholders access the same underlying data, the risk of exposing information to the wrong party is real. The mitigation is role-based access control with deterministic permissioning: each stakeholder role sees only the data fields and outputs they are authorized to see, enforced by system logic rather than user discipline. Data segregation must be designed into the architecture from the start, not added as a layer afterwards.

Workshop data gaming

Workshops required to submit structured data will, predictably, find ways to meet the letter of the requirement without meeting its intent — photos that technically exist but show nothing useful, estimates that are formatted correctly but contain inflated line items, milestone updates submitted in bulk after the fact. The mitigation is quality-weighted compliance scoring that assesses not just whether data was submitted, but whether it is accurate, timely, and internally consistent. Cross-referencing across data points is where AI adds real value in detecting gaming behavior.

Model drift

AI models are trained on historical data. As fraud patterns evolve, repair costs shift, and workshop behavior adapts, the models become less accurate. The mitigation is treating models as living systems, not finished products. Model versioning ensures you know which version produced which output. Performance monitoring tracks whether flag precision, triage accuracy, and benchmark relevance are holding or degrading. Scheduled recalibration keeps models current.

Change management failure

This is the most common failure mode and the least technical. The technology works, but the people and processes don't follow. Handlers revert to old habits because the new workflow feels slower or less familiar. Workshops disengage because the data requirements feel burdensome and nobody enforces them. Management loses patience because the benefits take longer than expected to materialize. The mitigation is phased rollout with demonstrated quick wins before scaling — and realistic timelines with interim metrics that show progress before the full trust stack is operational.

The difference between a trust stack that delivers and one that disappoints is almost always governance discipline, not technology capability.

‍

About the author

Frederik Bisbjerg is Co-founder and Managing Director of Axxion Claims Settlement Services LLC, the UAE's first dedicated motor claims third-party administrator, where he is building a compliance-by-design claims operating system with AI governance at its core.

His career spans more than two decades of insurance leadership across the MENA region, including roles as CEO of Al Wathba Insurance, Chief Transformation Officer at AXA Global Healthcare, Senior Vice President of Digital Transformation and Innovation at Daman National Health Insurance Company, and Executive Vice President at Qatar Insurance Group.

He also serves as Head of MENA and Digital Transformation specialist at The Digital Insurer, where he is a founding member of the world's first mini-MBA in Digital Insurance. Bisbjerg is the author of the best-selling Insurance_Next, a practical guide to transforming incumbent insurers into flexible, resilient organizations ready for the post-COVID, generative-AI era.

About Axxion Claims Settlement Services: Axxion is a Dubai-based end-to-end motor claims management company and the UAE's first dedicated motor TPA. Axxion is part of the Skelmore Group, a diversified automotive and insurance services group founded in Toronto in 1994, operating across North America and the Middle East with approximately $650 million in revenue and 4,000 employees.

‍

© 2026, Frederik Bisbjerg. March 2026.

‍