Q1. Could you start by giving us a brief overview of your professional background, particularly focusing on your expertise in the industry?

I’ve spent the past 27 years immersed in insurance and financial services, working across every part of the industry—from hands-on operations and transformation to delivery and advisory roles. My journey began at Genpact as a Black Belt, where I learned the ropes from the ground up. I then moved into operations leadership at Bank of America, before taking on full P&L responsibility for insurance delivery at WNS, overseeing portfolios ranging from $150 to $200 million. After that, I shifted my focus to building and leading consulting businesses—first as Vice President of Banking and Insurance Solutions at WNS, where I delivered annual contract values of $200 to $250 million, and then as Advisory Leader at Cognizant, leading a 150-person team across the UKI and Europe and growing our consulting revenues by over 10 million pounds. Today, I serve as an Insurance Consulting and Solutions Partner, where I head up strategic consulting for a $350 million portfolio, with a strong focus on driving AI-led growth.

My work bridges strategy and execution. Over the years, I’ve helped transform more than ten insurance carriers and broker ecosystems, consistently delivering 3 to 5 percent improvements in expense and loss ratios. For one European re/insurer, my efforts led to products hitting the market 20 percent faster and reduced exposure by about 30 million pounds a year. With a UK life and pensions carrier, I turned a negative NPS into a positive 32 and cut failure demand by 40 percent. In the Lloyd’s market, I led underwriting transformation that lowered the cost of doing business by 20 to 25 percent.

When it comes to AI, I’m not just an observer—I’m deeply involved in building real-world solutions. I’ve designed and delivered everything from agent-based underwriting platforms for the London Market to claims workbenches, premium audit automation, and regulatory remediation engines. I get hands-on with technologies like Anthropic, multi-agent systems, and retrieval-based models, and work right alongside core industry platforms like Guidewire, Duck Creek, Snowflake, and Databricks. My view is straightforward: technology without a clear strategy is just noise. AI is everywhere these days, but true strategic intent is rare. My work is about helping organizations bridge that gap.

Industry Outlook: Specialty and E&S, and the AI Imperative

My website: www.wedidntstartthefire.io

Specialty and excess and surplus (E&S) lines have become the main drivers of growth in global insurance. In the US, E&S premiums have seen double-digit growth for several years, as traditional carriers pull back from volatile risks and brokers move complex cases into the surplus lines market. Lloyd’s and the London Market remain hubs for these risks, and specialty capacity is also expanding across Europe and Asia-Pacific. But this rapid growth has outpaced the industry’s infrastructure. Many MGAs and specialty carriers earned their capacity based on relationships and underwriting expertise, but now, its reporting, data quality, and operational discipline that determine who keeps it. Put simply: today, survival for E&S MGAs depends on the quality of their reporting.

Right now, three powerful forces are converging on specialty markets all at once. First, capacity providers want machine-readable, auditable data on exposure, pricing, and claims—and they’re pulling capacity from partners who can’t deliver it. Second, rising loss costs in property, casualty, and natural catastrophe lines are squeezing margins, especially where data is weakest. Third, regulations are tightening: the EU AI Act, DORA, and conduct rules like the FCA’s are raising the stakes for anyone with weak controls. These challenges aren’t coming one after another—they’re hitting all at once.

This is where the AI imperative becomes concrete. Specialty underwriting has always been judgment-led, and it will stay that way. What AI changes is everything around the judgment: 
•    Submission intake
•    Enrichment
•    Triage
•    Exposure validation
•    Portfolio monitoring
•    Compliance evidence

From my own experience in the London Market, I’ve seen that underwriters spend a huge amount of time on tasks that aren’t really underwriting. Shifting that capacity back to risk selection is the single biggest economic opportunity in specialty insurance today. The carriers and MGAs that treat AI as a core part of their operating model, powered by their own proprietary data, will pull ahead. Those who see AI as just another tech purchase will end up with the same tools as everyone else—and little real advantage.

 

Q2. Where are you seeing AI create the greatest economic impact today, and which use cases could materially change insurers’ cost structures or profitability?

AI delivers the greatest impact when applied to core areas where insurers generate or lose value: underwriting selection, claims, and exposure validation. In more than ten carrier and broker transformations, I have seen that embedding AI in the operating model leads to 3 to 5 percent improvements in expense and loss ratios. These are the areas where profit and loss are most affected.

Underwriting Capacity

A significant portion of underwriter time is spent on administrative tasks instead of risk selection. AI-driven intake, enrichment, and triage return this capacity to underwriting judgment. For a mid-size commercial carrier, this can improve the combined ratio by 1 to 2 points.

Claims leakage and indemnity control

Commercial lines have faced years of underwriting losses—issues that pricing alone just can’t solve. AI-driven claims orchestration, leakage detection, and litigation prediction work directly on the loss ratio. For example, I’ve led programs that delivered about 10 million pounds in indemnity savings for a single carrier.

Distribution

Broker IQ

For years, brokers focused on buying renewal lists, expecting growth would come from selling more to the same clients. But in reality, no one has the time or resources to dive deep and find those extra opportunities.
Three out of four SME businesses are underinsured, but with low commissions, it’s just not practical for brokers to do detailed gap reviews, benchmark exposures, or check every missing line. What’s really needed is an AI-powered Broker IQ tool that gives a 360-degree view of the client portfolio and highlights the gaps.

Premium audit

North American commercial carriers lose billions every year to audit leakage. AI workbenches that match up exposure and payroll data can close this gap at a very low marginal cost—turning lost revenue straight into margin recovery.

Regulatory remediation

In the E&S admitted and non-admitted markets, MGAs are facing real penalties nowadays. Tax filing—different in every state—ends up draining time and energy that could be spent on growing the business.

In short, capacity may have fueled the growth of US insurance so far, but the real winners will be those who can prove compliance at machine speed.

The pattern is clear: true economic impact comes from using AI at decision points, not just handling documents. When insurers deploy AI where risk is priced, claims are settled, and exposure is verified, that’s when you see real change in cost structure. That’s exactly where I advise clients to focus their investments.

 

Q3. What changes when AI becomes part of an insurer’s core decision-making, and where could this create lasting competitive edge?

When AI becomes central to decision-making, three big shifts happen: decisions get faster, they become more consistent, and the data generated starts to build on itself—creating real, lasting advantage over time.

Decision speed

Underwriting and claims decisions that used to take days now happen in minutes, with each step fully evidenced and auditable. At a central UK institution, I helped create a unified, governed view of data that sped up decision-making by 40 percent.

Consistency

AI applies underwriting logic consistently across every submission, closing the gap between the best and average underwriters. This is especially important in specialty lines, where judgment quality really makes the difference.

Compounding data

Every decision feed the next, turning your own decision history—if organized well—into an asset no competitor can copy. Foundation models can be rented, but proprietary decision data is something you truly own.

Measurable outcomes

With AI, business results become visible at the transaction level. Management can adjust loss ratios and costs almost in real time, instead of waiting for the next quarterly review.

Accountability

AI-powered decision-making still needs real human ownership—not just of the outputs, but of the model’s behavior itself. Carriers who build this governance in from the start are moving faster, not slower, because both regulators and capacity providers trust their numbers.

Lasting competitive edge isn’t just about having the right AI model—it’s about the operating model and data foundation you build around it. In one project, this approach led to a 20 percent faster speed to market and saved about 30 million pounds a year in exposure. Carriers who treat every underwriting and claims decision as a chance to strengthen their own intelligence will steadily pull ahead over the next planning cycle.

 

Q4. How far will agentic AI realistically go in automating insurance workflows over the next 2 to 3 years, and which processes will shift from human-led to AI-led execution?

After 27 years of watching wave after wave of new technology hit this industry, I’ve stopped asking, “How far will AI go?” That’s not the right question—and it usually leads to the wrong kind of investments. The right question is: what does AI free our people to do? Carriers getting this right aren't measuring automation percentages. They are measuring how much more of their best judgment reaches the risks and customers that deserve it.

Looking ahead to the next 2 to 3 years, I see three big shifts coming:

Execution becomes machine work

Submission intake, triage, document handling, standard claims processing, compliance evidence—these were never where value was created. In fact, they’re the places where value used to leak away.

Judgment becomes empowered, not displaced

Now, the underwriter steps into a risk conversation with every relevant signal already pulled together. The claims handler enters a negotiation knowing the odds of litigation and the history of leakage. AI isn’t here to make the decision—it’s here to make sure the decision is more informed than ever before.

Accountability becomes the premium skill

As machines handle more of the routine, the people who remain are trusted with bigger, more consequential decisions. The industry won’t need fewer great people—it’ll need the same talented people but operating at a level that was never possible before.What that means:

Submission intake and triage

AI agents can now read, enrich, score, and route broker submissions end-to-end—just like my platform for the London Market does. The real gain isn’t just faster intake; it’s building a better selected book: fewer anti-selection losses and more underwriter time spent on the right risks.

Low-complexity claims

First notice, coverage verification, and reserve recommendations are now AI-led. The results? Lower indemnity leakage and shorter cycle times—which show up directly in the loss ratio and keep customers coming back.

Financial operations

Straight-through processing in areas like MGA premium payments can jump from about 40 percent to as high as 80 percent. That frees up capacity for growth and delivers the audit-ready accuracy that capacity providers expect.

Compliance execution

Evidence gathering and remediation calculation are machine work; I have seen AI compress months of data preparation into days. The outcome is regulatory exposure closed faster, at lower cost, with a defensible audit trail.

What will not move

Complex underwriting, claims negotiations, and regulatory sign-off will remain firmly in human hands for the next few years and probably beyond. That’s not a technology limitation—it’s because this is where margin, trust, and the license to operate truly live.
Here’s my bottom line: the insurers who come out ahead over the next three years won’t be the ones who automate the most. They’ll be the ones who realize, sooner than others, that AI’s true purpose in insurance isn’t to push human judgment out—it’s to let human expertise finally reach its full value. That’s the transformation worth leading.

Q5. How will the economics of outsourcing and managed services evolve as AI changes the operating model, and where is the largest structural reduction in cost-to-serve?

AI is fundamentally changing the way we think about outsourcing. Instead of simply moving tasks to lower-cost locations, automation is turning what used to be a transactional, labor-based approach into something truly transformational. Once a machine can perform a process step at almost no marginal cost, there’s no longer a need to chase lower wages. Based on my own experience running managed services businesses with $200 to $250 million in annual contract value, I believe we’ll see more change in the industry’s commercial model in the next five years than we have in the last fifteen.

From headcount to outcomes

The old model of charging by headcount is giving way to contracts focused on outcomes and shared gains. Soon, we’ll see commercial models where one price covers everything, driven by AI for each transaction or case. What clients are really buying isn’t the size of the delivery team anymore—it’s the provider’s AI capabilities and deep expertise.

Smaller, more skilled delivery teams

The question is shifting from 'which country should we outsource to?' to 'what’s the right mix of humans and machines?' Even before reaching full AI maturity, I’ve seen programs deliver 35 to 40 percent total cost reduction by redesigning around this principle.

Value moves to orchestration 

Providers who own their platforms, data infrastructure, and governance frameworks will be able to command premium prices. Those offering only generic capacity will see their prices steadily decline.

Business-level service measures

AI makes it possible to measure things like loss ratio impact, quote speed, and settlement accuracy at the transaction level. As a result, managed services contracts will increasingly be based on real business outcomes, not just operational service levels.

Largest cost-to-serve reduction

In areas like underwriting support, claims, and policy servicing—where 30 to 50 percent of the cost is tied up in fixing errors, rework, and document handling—I expect carriers who truly redesign these functions to see a 20 to 30 percent structural reduction in cost-to-serve over the next five years.

For carriers, brokers, and MGAs, this means it’s time to rethink and renegotiate outsourcing relationships—before the market resets. For providers, the focus should shift to investing in platforms and real outcomes, not just more capacity. The structural savings are absolutely real, but they’ll only go to those who put AI at the core of their operating model, not to those who simply bolt it onto the old way of working.

Q6. After 25+ years of transformation, what is the one lesson every insurance CEO and investor should keep in mind as AI reshapes the industry?

If there’s one lesson I’ve learned in 27 years, it’s this: technology without a clear purpose is just noise. AI is everywhere, but true strategic intent is much harder to find. In insurance, that gap is exactly where value is created—or lost.

Strategy first

Every wave I’ve seen—from core platforms to digital to now AI—has rewarded the carriers who started with clear strategic questions. They turned technology into real improvements in their ratios. The ones who started with the tools themselves usually ended up with a pile of pilots and a lot of disappointment.

Sequence matters

When transformation is done right, business results follow naturally. Start by building a solid data foundation, then focus on redesigning workflows, and only then move to agentic execution. If you skip the groundwork, you’re basically building on sand.

Understand token economics before you scale

AI isn’t free—each agent call, document processed, or reasoning loop costs money, and those costs add up quickly at scale. I’ve seen business cases for AI that looked great on paper but fell apart in production, simply because no one calculated the cost per decision against the value. The fix is simple: know your cost-to-serve per AI transaction just as well as your expense ratio, and match the model size to the task. Most insurance workflows run just fine on smaller, more affordable models, smartly chained together.

Use AI for the right reasons

AI should be used where it genuinely improves decisions, protects customers, or cuts costs without sacrificing value. It doesn’t belong in flashy press releases or in transformation programs that are just for show. My test with every client is simple: if this AI vanished tomorrow, would any real outcome suffer? If not, it shouldn’t have been funded in the first place.

Watch the moat, not the demo

Any competitor can rent the same AI models. The real, lasting value comes from your own decision data, strong governance, and the operating model you build around the technology.

Time is the constraint

There’s a two-year window to become a truly AI-enabled insurer—not five. Boards need to allocate capital and management focus with that timeline in mind.

For investors

Investors should look for management teams willing to change how the business actually runs—not just those who can talk about AI. The real question isn’t “What’s your AI strategy?” but “Which human-led processes have you already committed to machine execution, and who’s accountable for each one?”

When you get this right, the results are real: double-digit gains in speed, single-digit improvements in combined ratio, and cost reductions in the tens of millions. Get it wrong, and you end up with two years of pilots, an exhausted team, and a competitor who beat you to it. The technology is ready—the question is, is your leadership?