For some reason, most Americans seem to think that when the U.S.-Iran conflict comes to an end, oil prices and the broader economy will quickly bounce back to normal. Unfortunately, that is just not realistic, and the longer-term damage is already set in motion. Subject matter experts are predicting a 12- to 18-month correction period once the situation stabilizes. The backup of oil tankers in the Strait of Hormuz will take at least a year to clear.

A year‑long oil crisis would hit both automobile sales and auto insurance in ways that go far beyond just higher gas prices. The short version: vehicle demand would likely shift sharply toward fuel‑efficient and electric models, overall sales could soften, and auto insurance costs would almost certainly rise due to inflation, repair costs, and economic stress. Below is a structured breakdown grounded in recent reporting and economic analysis.

Impact on Automobile Sales

Demand will shift toward fuel‑efficient and electric vehicles. When fuel becomes expensive for a long period, consumers rethink what they drive. Economic theory treats vehicles and gasoline as complementary goods, meaning high fuel prices suppress demand for gas‑heavy vehicles. Buyers tend to move away from trucks and large SUVs and toward smaller, more efficient cars or EVs.

Overall auto sales could decline. A prolonged oil crisis raises household expenses across the board. With budgets squeezed, many consumers delay big purchases like cars. This effect is amplified if the crisis also disrupts supply chains or raises production costs—both of which are likely when oil prices stay high for months.

Higher vehicle prices due to supply chain strain. Geopolitical disruptions tied to oil crises often spill into shipping and parts availability. Recent reporting shows that conflicts affecting oil supply also cause shipping delays, higher transport costs, and production cuts by major automakers. Toyota, for example, has already reduced output in response to Middle East instability. Fewer cars produced means higher prices for both new and used vehicles, further dampening sales.

Impact on Auto Insurance

Rising premiums driven by inflation and repair costs. Auto insurers are already facing a "severity crisis": repair costs have surged due to inflation, supply chain issues, and the increasing complexity of modern vehicles. A prolonged oil crisis would worsen these pressures by raising transportation and parts costs. Insurers have been "racing to take rate," and pessimistic outlooks suggest continued premium increases.

Higher replacement costs due to vehicle shortages. If automakers produce fewer vehicles because of high energy costs or supply disruptions, replacement vehicles become more expensive. Insurers must pay more for totaled cars, which pushes premiums higher. This dynamic has already been observed during labor strikes and supply chain disruptions.

Changes in customer retention because of Increased financial stress. When households face sustained high fuel costs, they may struggle to keep up with insurance payments. Analysts warn that squeezed budgets can lead to policy lapses, reduced coverage levels, or shopping for cheaper (and sometimes inadequate) policies.

More accidents in stressed industries. In sectors tied to oil and gas, worker shortages and fatigue have historically increased accident rates, which in turn raise liability claims and insurance costs. While this is industry‑specific, it contributes to overall market pressure.

The Big Picture

If the oil crisis lasts a year or more, the most likely outcome is:

• Automobile sales soften overall, with a strong shift toward efficient and electric models.
• Large SUVs and trucks lose market share, unless essential for work.
• Vehicle prices rise due to supply chain strain and higher transport costs.
• Auto insurance premiums continue climbing, driven by inflation, repair costs, and higher replacement values.
• Consumers face financial strain, leading to more lapses, reduced coverage, and slower sales cycles.

Reality bites, but understanding these outcomes and challenges will enable all participants to plan and adjust accordingly.

Paul Carroll

Denise, we were talking the other day about the fundamental changes occurring in insurance, and you had quite a list. Could you start us off by walking us through some of those?

Denise Garth

The industry is changing a lot, and it's not just technology — it's everything. Risk is changing, customer demographics and expectations are changing, where people are living is changing. 

One of the biggest things we're seeing is the growing protection gap. The cost of insurance has increased significantly due to climate and weather events, rising claims costs, and the legal challenges the industry faces. It is unsustainable for customers, forcing them to make difficult decisions such as not buying insurance, switching for a lower cost, increasing deductibles, and more. It is a tipping point of change.   

We see a new era for insurance — one that's really built around intelligence to enable adaptability.

The way forward is going to require both an operating model and a technology foundation redesign and redefinition. We've been talking about transformation for the last 10 to 20 years, and in most cases, it was about ripping out the technology and putting in something new over the existing operating model. Now we must rethink the operating model: how we want and need to do business to remain relevant.

In today's world, products are evolving. You still need auto, but there are so many variations of it now — autonomous vehicles, people doing things with Uber and the gig economy. There's a whole different set of product types needed to support those, and that goes across all products, whether it's P&C or L&A&H. 

We have to do business in a way that fits this future, not the past.

Our operating models have been crafted over decades around a myriad of constraints, business assumptions, and challenges from the past. They've evolved by layering in technologies, manual work, point solutions — and we now face what I call a "spaghetti infrastructure" that has created a really inefficient, unprofitable, and employee-constrained operation. It's added a level of complexity on top of an already complex business. 

Instead of just replacing technology with the next modern core solution, we have to think about what it is we compete on. That's where technology really begins to come into play — not just cloud-native technology and robust core systems, but now AI, both in terms of technology infrastructure and business architecture that can redefine the operating model and business processes. 

In a webinar I just did, I shared that 82% indicate they want to do something with AI, but very few are actually doing it, or they're doing it in a piecemeal way. AI needs to be more than just an add-on technology. It has to be embedded into and redefine how we do business, so you can constantly optimize what you're doing. That redefines the overall business value of cloud and AI-native core that the market begins to see and realize in business outcomes.

I predicted that by 2030, we could see a 20-point reduction in expense ratios — and it's starting to happen as you see publicly traded insurers talk about what they're doing with AI. That is going to completely change the competitive landscape. 

Paul Carroll

For me, the big thing I see companies potentially missing — because I've seen them miss it in other waves of technology over the past several decades — is the need for the agility you mention.

Gen AI is going to allow the sort of breakthrough that Amazon produced in the first wave of the internet. It didn’t just do the old things better; Amazon reinvented retail. If insurers lock themselves into developing a better form of what they've done before, they're going to miss out on a lot of opportunities.

From a technology standpoint, how do you enable the agility that insurers need?

Manish Shah

Before diving into the solution, I want to make sure we also look at the broader, common theme underlying these problems. A lot of people blame the insurance industry for not having modern systems, for not knowing their customers, for not having the right products or pricing. But if you really dig deep, the biggest issue facing the insurance industry — the one causing all those other problems — is that it simply cannot keep up with how fast the world is changing. Insurance is out of phase.

Customer expectations are significantly different and changing almost daily. There’s a huge change in risks and in how those risk profiles are developing. And the technological advancements happening today are leaps and bounds faster than what insurance companies' general culture allows them to absorb.

They're not unaware of the problem. The issue is how fast they can adopt new technology, how fast they can change their culture and get to changes in products, better pricing, better distribution, and so forth.. 

Our view is that it's not just about using technology or solving a niche problem. It's about making your mission-critical systems nimbler and relying on a partner and ecosystem framework rather than a traditional command-and-control framework. 

Not every innovation has to be built in-house from the ground up. The real value companies can leverage is to test the technological innovations that companies like ours bring to them in a meaningful way — roll them out to customers, learn from them, test them, understand user behavior, and refine them.

That's why our approach is not simply about selling technology or a core system. It's about having intelligence built into every workflow, every process, every customer interaction — so you can get meaningful feedback from customers that allows you to evolve faster than the rest.

It's not a technology discussion — it's a speed discussion. How fast can I validate my ideas? That, clearly, is the biggest impediment in the industry.

Most people are still grossly underestimating what AI can and will do to every single business. Insurance is not an exception. Regulations will shield you only for so long, but when it comes to customer service, operational efficiency, improved profitability, faster turnaround, claims resolution, and better underwriting — AI, and more importantly, agentic AI, is going to play a huge role in every single one of those areas.

Whether people embrace it or resist it, in the next 18 to 24 months, a hybrid workforce — built with humans and AI agents working together — is going to be common. We're literally talking about leveraging artificial intelligence not as a tool but as an entity that works alongside humans. And that means the human workforce is going to have a very different role. They won't be writing the first draft — they'll be validating it. That is a huge cultural shift.

If organizations don't start engaging with this thought process early and experimenting with it now, they'll eventually be pressured to do it in a hurry. And if you try to implement this in a rush, even if you can get the technology in place, you cannot simultaneously implement the cultural shift that needs to accompany it. Doing it sooner is critically important.

Denise Garth

We talk about the "capacity gap." The capacity to have the right type of people running the business inside an insurance company is under significant strain — particularly given that a large percentage of the workforce is expected to retire by 2030. Estimates put those losses at 40% to 50%. You're going to lose your underwriters, your claims adjusters, your billing professionals — people who know your legacy systems, let alone people who understand your products and your business.

That's exactly where the hybrid workforce comes into play. Not only can it help you do more with the resources you have, but it can also educate and train new people in a consistent way — creating real value, consistency, and quality for those coming in and trying to learn this business. It gives them the confidence to do the work and learn along the way. That's a major factor in all of this that a lot of insurers haven't fully faced up to yet.

Paul Carroll

Peter Drucker used to say that culture eats strategy for breakfast. And when you look at AI — or just the new technology environment, in general — if you approach it as a destination, something you're going to do once, you're going to fail. 

It has to be a cultural shift, something you work on this week, next week, next month, and the month after that. 

Denise Garth

It really comes down to leadership, because you're going to have to redefine the organization and people's roles — jobs are going to look very different. 

Paul Carroll

How does software need to evolve to support a hybrid workforce of both humans and AI agents?

Manish Shah

Today’s software was designed to be used 100% by humans. And human users have a little bit different constraints than AI users. For example, humans can't process too much information at once. We need multipage forms in a user interface, relational databases, more structured data — things like that. AI agents don't have those same constraints. Software today must be designed for both people and AI agents to do the work they’re best suited for. 

Toward the latter part of the year, we plan to release a brand-new user interface, suited for each type of user. Providing seamless handoffs between them is also a key part of that design consideration. 

The current core system user design is simply not going to be adequate for where the world is moving. The industry has come a long way in the last 20 to 25 years in modernizing, but the fundamental pain points are still there — how long it takes to implement modern software, the cost, how long it takes to maintain it, the total cost of ownership. 

Just like Claude has created a significant dent — in a lot of people's minds and in the markets — with the idea that "I can build the software," we think the same kind of shift is possible for enterprise implementation. Sure, that came with a lot more enthusiasm than realism at first, but I think it will get there.

Why can't AI implement our software? Why does an implementation take three years? Our goal is to build a Claude-like AI capability that interacts directly with business users and translates that into system configurations — allowing our customers to actually move forward.

Paul Carroll

Thanks, Denise and Manish. 

About Denise Garth

Chief Strategy Officer at Majesco, Denise Garth drives thought leadership and innovation strategy for insurers worldwide. She’s a global voice on digital transformation, customer experience, and the future of intelligent insurance ecosystems, shaping how carriers modernize and reimagine their business models.

About Manish Shah

President and Chief Product Officer at Majesco, Manish leads global product innovation across intelligent core systems, AI-powered platforms, and digital ecosystems. A visionary technologist, he’s known for helping insurers accelerate modernization while staying true to human-centric design and trust.

For decades, insurance has relied on a model that assumes time is on its side. Risk could be assessed, priced, and adjusted in cycles. Products evolved gradually, and systems were built for control rather than speed. That model is now under pressure in ways it was never designed to handle.

The issue is not that insurers lack insight. Most organizations have more data than ever before, along with increasingly sophisticated models to interpret it. The problem is far more practical: they cannot act on that insight fast enough. Pricing updates remain tied to fixed cycles, model changes take time to deploy, and by the time adjustments are implemented, the underlying risk has already shifted.

Inside insurance organizations, this tension is well understood. There is no shortage of awareness or intent. The frustration comes from the gap between what teams know needs to happen and what they can execute. Pricing changes sit in queues, model updates wait for deployment windows, and while those changes move through the system, the underlying risk continues to move.

The gap between the speed of risk and the speed of response is no longer just inefficiency. It's showing up in loss ratios, missed growth opportunities, and an increasing inability to compete on speed.

A model that cannot keep up

Insurance was not designed for continuous change. Pricing is still adjusted at defined intervals, underwriting models are updated periodically, and product changes move through systems that assume a relatively stable environment.

Risk no longer behaves that way. Exposure can shift materially between pricing reviews. New data arrives continuously, often from sources that did not exist even a few years ago. By the time updates are implemented, the assumptions they were based on are frequently out of date.

Most insurers recognize this dynamic. The challenge is not diagnosing the problem, but overcoming the structural constraints that prevent them from responding in real time. Legacy systems, internal processes, and the way decision-making is organized all introduce delay, even when the business is trying to move faster.

The result is a fundamental mismatch between how risk evolves and how insurance operates.

When technology slows you down

Much of the industry conversation around innovation focuses on adopting new technologies. But for many insurers, the more immediate issue is the technology already in place.

Core systems continue to underpin underwriting, pricing, and product configuration, yet were built for a different era. They prioritize stability and control, which made sense when change was incremental, but they are far less suited to an environment where conditions shift constantly.

This creates a form of operational inertia. Even relatively straightforward changes can trigger complex processes, requiring coordination across multiple teams and systems. As a result, external changes move faster than internal responses. Updates queue behind IT backlogs, implementation timelines stretch, and opportunities to respond to emerging risks are missed.

It's not a lack of capability that holds insurers back. It's the difficulty of translating that capability into action within the constraints of the existing operating model.

The AI gap is an execution gap

The same pattern is playing out with AI and advanced analytics. The potential is widely understood, and in many cases, already proven. More precise pricing, improved risk selection, and better customer engagement are all achievable outcomes.

What remains unresolved is how to operationalize those capabilities at scale.

In many organizations, AI is still being deployed as a series of point solutions rather than integrated into the core of decision-making. Data remains fragmented, insights are generated in isolation, and the process of moving from analysis to action is slower than it needs to be. This is not a failure of ambition but one of integration.

Without an operating model that can absorb and act on these capabilities continuously, AI risks adding another layer of complexity rather than delivering meaningful transformation. The gap between what is technically possible and what is practically achievable continues to grow.

Innovation that arrives too late

One of the clearest consequences of this dynamic is the speed of product innovation. Emerging risks require new forms of coverage, more flexible pricing, and the ability to adapt offerings as conditions change. Yet bringing new products to market remains a slow, resource-intensive process. By the time a product is launched, the risk it was designed to address may already have evolved.

In effect, insurers are often pricing yesterday's risk in today's market.

This lag has direct commercial implications. It limits the ability to seize new opportunities, exposes reliance on outdated assumptions, and makes it harder to compete in areas where speed and adaptability are becoming critical.

More than an efficiency problem

It's tempting to frame these challenges as operational inefficiencies. At its core, this is a question of missed opportunity. Every delay in responding to changing risk conditions shows up somewhere. In pricing that no longer reflects exposure. In products that reach the market too late. In capital deployed against assumptions that are already outdated.

Over time, this erodes both profitability and competitiveness. It also has wider implications for the role insurance plays in the economy. When insurers cannot respond quickly enough to evolving risk, it becomes harder to price and transfer that risk effectively, which in turn affects how capital is deployed.

A breaking point for the operating model

The insurance industry has adapted to change many times before, but the current moment is different in both speed and scale. What the industry is facing is not a series of isolated challenges, but a structural shift in how risk behaves. The operating model that has supported insurance for decades is reaching its limits.

Closing the gap between the speed of risk and the speed of response will require more than incremental improvement. It will require a fundamentally different approach, one that allows insurers to move from periodic decision-making to continuous, real-time action.

The industry is not short on data, insight, or ambition. What it lacks is the ability to translate those strengths into action at the pace the market now demands. That is why this moment feels different. This is not simply another innovation "phase," it's the point at which the traditional operating model breaks.

Small commercial construction has a coverage-gap problem, and the industry still tends to frame it too narrowly. This is not simply an education issue. It is a systems issue spanning intake, underwriting, product selection, agency guidance, renewal, and digital distribution.

Most businesses in the small-contractor world have at least one real coverage gap, and usually more than one. The same hot spots show up repeatedly. Subcontractors are a big one: no certificates on file, subcontractors carrying skinny limits, or general liability policies that quietly carve out subcontractor work or action-over claims. Tools, equipment, and materials are another weak link. Without good inland marine coverage, anything not bolted down or specifically listed, including tools in trucks, on jobs, or in storage, is basically uninsured. Completed operations are often thin, even though many claims show up months after the job is done. Many small contractors also remain bare on EPL, cyber, and professional liability, even when they are doing design-build or heavy advisory work.

That pattern persists because the market still treats small contractors like downsized versions of big construction firms, when they behave more like volatile micro-businesses. They pivot fast. A three-person general contractor may effectively be running a 40-person operation through subcontractors, while underwriting is still staring at W-2 payroll. The issue is not a lack of products. It's a lack of connectivity between what these businesses actually do and how the insurance workflow captures risk.

Where the mismatch shows up most often

General liability is still the big category that contractors misunderstand. Many assume it covers everything, including design mistakes, employee issues, and their own stuff. In reality, it usually does not include professional liability, EPL, or personal property, and it may limit or exclude some subcontractor work. Inland marine or tools coverage is often confused with GL or property, or skipped entirely, which means tools in trucks, on sites, or moving between jobs are underinsured or not insured at all.

Workers' compensation also gets dodgy when owners try to call everyone a 1099 to save premium, then find out the hard way they have misclassified people and have no real coverage for injured subcontractors. Builders' risk is often assumed to be baked into GL or the owner's policy, so ground-up jobs and major renovations go forward with no project-specific property coverage.

These gaps usually start at the front door and then get locked in at every step. Intake is rushed, so applications understate revenue, gloss over higher-hazard work, and skip key details like subcontractors, storage locations, or any design role. Underwriting on small accounts leans too heavily on class codes and checkboxes instead of actually looking at job mix and contracts. Agencies under pressure to be fast and cheap default to contract minimums, not coverage that matches how the contractor actually operates. Once the account is bound, renewals become copy-paste unless a new job contract or pricing issue forces changes.

But a small premium does not mean simple risk. If intake captures only a partial version of the business, every downstream step becomes a more efficient way of institutionalizing the wrong answer.

Where data and AI can genuinely help

The industry talks a great deal about data and AI closing protection gaps. In small commercial construction, those tools can absolutely help, as long as there is honesty about their limits.

They are strong at spotting mismatches. A business described as a handyman operation may suddenly show structural steel on its website, permits, or social footprint. Revenue, payroll, job types, and contract requirements may suggest limits that look very different from what was initially requested. AI and external data can also prompt for missing pieces that matter, including subcontracts, storage locations, equipment schedules, and certificates of insurance.

But those tools still fall short when the intake data is thin or wrong because that just becomes fancy math on bad inputs. They also struggle with nuance in construction contracts, indemnity wording, project delivery methods, and all the gray areas humans still argue about. When a contractor reinvents the business every six to 12 months without clear signals, the models lag reality.

AI is most valuable here as a signal-detection layer that surfaces where the representation of risk and the operating reality have drifted apart. That signal layer is powerful, but it still needs a human who understands construction and coverage to translate signals into real decisions at the account level.

Why digital-first works only part of the time

Embedded insurance and digital-first distribution can work for simple, low-hazard contractor risks. Solo trades with straightforward work, such as interior painting, basic handyman services, and simple flooring, are often a good fit for one-click experiences, especially for certificates of insurance and small endorsements.

But once the risk includes heavy use of subcontractors, multi-story or structural work, design-assist, unusual materials, or specialized sites such as data centers and hospitals, the account moves out of click-and-bind territory. A more consultative approach becomes necessary because someone has to ask harder questions about scope, contract language, and jobsite conditions.

The opportunity is not to choose between human expertise and digital efficiency. It's to build a true human-plus-machine model, with AI surfacing the right questions and construction-savvy agents and underwriters interpreting the messy reality on the ground.

What needs to change

To make small commercial construction more consistent and scalable for carriers, underwriting, data access, and product design have to be rethought together.

It starts with smarter intake: third-party data such as permits, licensing, online footprint, and certificate tracking, combined with a tight set of questions about subcontractors, heights, structural work, and design responsibility.

On the product side, the market needs to move away from one-size-fits-all BOPs and toward modular contractor stacks: general liability with completed operations, inland marine, builders' risk, professional liability, EPL, and cyber. The industry should also standardize how it handles subcontractor injury, action-over exposure, and unscheduled subcontractors instead of burying giant gaps in endorsements no one reads.

Renewal is another missed opportunity. Claims data and operational signals should trigger smarter follow-up questions. Did the insured add design services, go higher, start public work, or pick up data-center jobs? That is how a tiny general contractor winning a slice of a larger data-center build gets structured correctly instead of being shoved into a tiny-limit package or declined outright.

What a better ecosystem looks like

In a healthier ecosystem, small contractors would not be walking around with Swiss-cheese policies that only reveal the holes when a lawyer gets involved. Coverage would adapt as work changes, with limits and modules responding to live data instead of stale applications. Policies and quotes would spell out major exclusions, including subcontractors, height, professional services, and residential work, in plain language right up front.

Subcontractors would be run through shared platforms for vetting, certificates of insurance, and standard hold-harmless language, feeding cleaner data straight into underwriting. Intake, underwriting, and renewal would function as a continuous risk-monitoring process.

That is the real innovation challenge in small commercial construction. The market does not need a watered-down version of large-account insurance. It needs a more adaptive, reality-based version of small-account insurance built for businesses that change faster than the forms designed to cover them.

Artificial intelligence has moved from experimentation to active deployment across the insurance value chain. Predictive models are augmenting underwriting, generative AI is accelerating claims and policy servicing, and agentic systems are beginning to coordinate multi-step workflows that previously required human handoffs.

Carriers that move AI into production consistently address three requirements: a data foundation built for AI consumption, an architecture engineered for production conditions, and an experience layer designed for the people who use it.

The data foundation

Connecting data for AI consumption is the first major engineering effort in any serious AI program. Policy systems, claims platforms, loss history, external feeds, and regulatory data have accumulated in separate architectures over decades. Each was built to serve a specific function, and connecting them for AI requires deliberate work. The design choices made at this stage carry forward into every subsequent AI operation.

Where data lives determines the cost and compliance profile of those operations. Running inference inside a governed platform already equipped with access controls, audit logging, and encryption carries lower compliance exposure and lower per-operation cost than routing data to an external model API. That decision is made early and is expensive to reverse.

The highest-value early work in most programs is automation and data engineering. Normalizing loss runs, structuring adjuster notes, and building a reliable integrated view of a risk generate analytical value before any model is involved. These steps build a foundation that extends to subsequent use cases without being rebuilt each time.

Production-ready AI architecture

A production-ready architecture must do three things well: control what runs, make it run reliably at scale, and provide clear visibility into whether it is performing as expected. These elements depend on one another.

First, every agent in production must be pinned to a specific, documented model version. A change to an agent's instructions carries the same functional impact as a change to a model's parameters. Both require the same change management controls. The compute engine decision — in-warehouse versus external — determines data residency, latency, cost, and compliance exposure. That routing decision should be explicit, documented, and revisable.

Second, the system must handle real production conditions. Inference at scale requires prompt caching, token quotas, and cost attribution by agent, use case, and business unit designed in from the start. At peak underwriting volumes or during catastrophe response, uncapped spending quickly becomes a budget event. Long-running tasks need async processing and checkpointing so they can resume cleanly after interruptions. Failure handling — dead letter queues, retry logic with backoff, and idempotency — must be part of the original design so transient outages do not become analyst problems.

Third, the architecture must make performance visible. Override rates serve as the leading indicator of model quality in production. When underwriters or adjusters consistently modify AI outputs, something has shifted in the model, the data, or the business context. Distributed tracing with shared correlation identifiers across every service call turns failure diagnosis from a reconstruction exercise into a lookup. Every AI decision must be recorded with its inputs, agent version, model parameters, and any human override so that when a regulator asks how a specific underwriting decision was made, the answer is already waiting in the log.

The experience layer

Platform selection shapes what provenance is even possible, while the interaction pattern determines how that provenance reaches the user. For work involving policy data, medical information, or PII, the delivery platform must keep that data within the appropriate governed perimeter. The compliance exposure from getting this wrong surfaces at examination time.

The interaction pattern should match how the work actually gets done. Conversational interfaces suit knowledge retrieval. Structured outputs suit decisions feeding downstream systems. Embedded AI integrated into the application an underwriter or adjuster already uses suits workflows where adoption depends on minimizing context-switching.

Underwriters and claims professionals acting on AI-generated outputs need to know what data the output was based on and whether a human reviewed it. Without provenance, usage patterns split between over-reliance and skepticism.

Starting the journey

Build each component for the production environment from the first use case. Size the data foundation so it can extend beyond the initial project. Design the architecture for the actual load it will face, with observability and failure handling included from day one. Shape the experience layer around the real workflows of underwriters and adjusters, using platforms that already satisfy the compliance requirements of the data involved.

More than $12 billion worth of insurance M&A deals have been announced so far this year, up from $10 billion last year, demonstrating the importance of M&A as a strategic tool. For carriers, brokers, and technology providers alike, M&A can accelerate growth, expand capabilities, and reposition firms for an increasingly digital and data-driven marketplace. Yet experience has shown that while deals are relatively easy to announce, they are far more difficult to execute successfully.

ACORD's recent Carrier Mergers & Acquisitions study sought to better understand both deal rationale and – more importantly – the implications and imperatives that drive successful outcomes. The study revealed clear differences in both the prevalence and effectiveness of carrier M&A strategies across four distinct deal rationales.

Diversification is an attempt to expand the portfolio by acquiring new revenue and earning sources. It accounted for 41% of carrier deals and delivered strong post‑transaction returns (+14%), making it the most prevalent and one of the most effective drivers of insurance M&A.

Core Expansion increases share across areas in which the insurer already executes, such as products, geographies, channels, and customer segments. These transactions represented 29% of deals and generally produced positive outcomes when valuation discipline, relationship preservation, and integration execution were well managed.

Scale & Scope goals include amortizing fixed costs and improving resource access by increasing absolute size, or expanding scope across strategic and tactical dimensions. These deals ranked third in frequency but were the only category to generate negative returns (‑14%), reflecting the risk of overstated synergies, underestimated integration risk, and diseconomies from added complexity.

Capability Acquisitions intend to optimize the risk, cost, and time associated with developing new or enhanced internal capabilities. While only 6% of transactions, these generated the highest returns (+28%) by closing strategic gaps and improving competitiveness rather than relying on scale alone.

Post-Merger Execution: Risks and Imperatives

Analysis of post‑transaction performance shows that outcomes are shaped far more by integration execution than by deal rationale, creating a clear divide between successful and underperforming transactions. Across the more successful transactions, acquirers focused on converting the deal thesis into a small number of high‑impact initiatives with clear ownership, targets, and governance under a single integration authority.

Early leadership and cultural alignment, protection of core operations, a defined target operating model and technology roadmap, and proactive talent retention were critical. The best performers track synergies through a single source of truth, communicate consistently, and sequence integration to deliver near‑term, no‑regret value before more complex change.

For transactions that fell short, value destruction was driven primarily by execution, not deal logic. Cultural friction, leadership misalignment, and technology integration complexity routinely slow decision‑making, erode talent, and dilute expected scale and efficiency benefits – risks amplified by regulatory burden and integration fatigue. Without disciplined value‑capture management, synergies identified in diligence often dissipate during integration.

The Role of Standards in Post‑Merger Integration

Standards are a critical execution lever in post‑merger integration, particularly in data‑intensive and highly regulated insurance environments. Industry data standards provide a foundation for reducing integration risk and accelerating value capture by addressing common sources of post‑deal friction. In many transactions, value leakage stems not from flawed strategy but from inconsistent data definitions, fragmented messaging, and limited interoperability across legacy and acquired systems. Common standards mitigate these challenges by establishing a shared insurance data language across underwriting, claims, policy administration, billing, reinsurance, and finance.

In the near term, industry standards enable effective "bridge integration," allowing disparate systems to exchange standardized data and messages without immediate core system replacement. This accelerates the consolidation of operational and financial reporting, enabling leadership to establish a credible single source of truth for synergy tracking, performance management, and regulatory oversight. Standardized data also reduces reconciliation effort, manual workarounds, and control gaps that frequently obscure results and slow integration progress.

Common data standards support several high‑value integration use cases across the insurance value chain. In underwriting and product management, common data models improve portfolio visibility and enable faster rationalization across products, lines, and geographies. In claims and servicing, messaging standards support more consistent customer and distributor experiences during integration, reducing disruption and service degradation. In reinsurance, finance, and risk management, standardized data structures enhance exposure aggregation, capital reporting, and regulatory compliance across the combined enterprise.

Over the longer term, standards-based architectures provide a stable foundation for phased modernization and system rationalization. By reducing reliance on bespoke point‑to‑point integrations, standards lower cost, improve flexibility, and shorten time‑to‑value for future acquisitions.

As insurance M&A accelerates and transactions grow larger and more complex, post‑merger execution – not deal ambition – will continue to drive shareholder value. Leading organizations will distinguish themselves by treating integration as a strategic capability, embedding discipline, governance, and data alignment from the outset. Industry data standards enable speed, control, and transparency while preserving future optionality. Positioned correctly, standards help protect the franchise, manage execution risk, and sustain value creation in a data‑driven insurance industry.

TLDR: Deploying agentic AI without governance infrastructure accumulates regulatory and operational exposure faster than most carriers recognize. Carriers scaling with confidence are those that made governance foundational and not remedial.

The Governance Infrastructure Insurers Need

Agentic AI is taking on consequential decisions across underwriting and claims at unprecedented speed and scale. The governance infrastructure at most insurers has not kept pace.

A simple prompt change - a few lines of text updated in a configuration file - can alter how an AI agent reasons about risk across every submission it processes. In a traditional predictive model, the equivalent change requires a full retraining cycle - weeks of documented work, validation runs, and a formal change record. In an agentic system without governance infrastructure, the same functional impact happens with no audit trail. Regulatory and operational exposure accumulates as a result.

Organizations scaling agentic AI with confidence have recognized this gap early and built the infrastructure to close it - because the ability to answer basic questions about their AI systems is a precondition for operating them responsibly at scale.

Why Agentic AI Strains Traditional Model Governance

Insurance organizations have spent years building model risk management capability for predictive AI - pricing models, fraud scores, and reserve estimates. These systems are well understood in governance terms. They take defined inputs, apply learned parameters, and produce a single output that a human then acts on.

Agentic systems change this structure in fundamental ways. Reasoning chains span multiple steps - querying data, evaluating evidence, calling external tools, forming intermediate conclusions - before producing a final result. A simple input-output log is no longer adequate.

Prompts are the governing parameters for AI agents. Changing a system prompt is functionally equivalent to changing a model's weights. Tool calls are data transactions that can potentially send confidential information to third-party systems. Human oversight placed only at the end of the workflow misses the dozens of consequential micro-decisions made along the way.

What Regulators Are Expecting

SR 11-7 is the de facto governance baseline. State insurance regulators are applying its principles - model inventory, independent validation, change management documentation, continuing monitoring - in examination practice. Any carrier deploying AI that influences underwriting or claims decisions should treat SR 11-7 as the minimum standard.

The NAIC AI Model Bulletin's eight principles are now shaping market conduct exams. Explainability carries the sharpest operational bite: if an AI system influenced a coverage decline, the insurer must be able to explain why in specific, contemporaneous terms. Colorado SB21-169 adds testing for proxy discrimination, documentation of external data sources, and annual certification. Similar requirements are advancing in California, New York, and elsewhere.

One pressure that did not exist two years ago is now concrete - carriers without documented AI governance frameworks are facing coverage exclusions and premium increases on their own AI liability policies. The industry that applies governance scrutiny to its insureds is now applying it to itself.

Governance Capabilities

A governance framework for agentic AI requires six distinct capabilities. Each addresses a specific gap in how these systems are built, operated and governed.

1. Asset registry - Answers the first question regulators ask: what AI is in production, who owns it, and what version is live. Every agent, task, prompt, and tool is stored as a versioned database record - not hidden in code. SR 11-7's model inventory requirement and the NAIC transparency principle both resolve to this capability.
    
2. Lifecycle framework - Enforces the change management discipline that agentic systems otherwise lack. Every asset version moves through a defined sequence of states - from draft through shadow deployment to champion - with human approval gates at the points that matter. A prompt change cannot reach production without the same controls applied to a code change.
    
3. Testing & validation - Replaces ad-hoc demonstration with structured evidence. Before any agent version reaches production, it is tested against ground-truth datasets labeled by subject-matter experts, producing precision, recall, and fairness metrics stored against the specific version. Colorado SB21-169's bias testing requirement and SR 11-7's independent validation requirement both have a direct answer here.
    
4. Execution control - Ensures that what runs in production is exactly what governance approved. At runtime the framework pulls the current champion configuration from the governed registry. Agents can access only authorized tools and approved parameters. Governance is enforced at execution, not assumed after the fact.
    
5. Decision logging - Produces the contemporaneous record that the NAIC explainability requirement and Colorado's adverse-action provisions demand. Every AI invocation is logged with the exact prompt version, inference parameters, tool calls, inputs, and output. When a market conduct examiner asks how a specific decision was made, the answer is a query, not a reconstruction.
    
6. Compliance & reporting - Makes the governance data useful to the people who need it. Model inventory reports, model cards, override-rate analysis, and regulatory evidence packages are generated on demand from the records the other five components accumulate, not assembled manually when the examination notice arrives.

The Infrastructure Argument

Policy documents alone cannot operationalize these capabilities. A policy requiring documented approval for all agent changes creates the obligation but not the mechanism. Under operational pressure, informal processes prevail.

The insurance industry has already solved an analogous problem in actuarial pricing systems. Algorithms are versioned, changes require documented approval, prior versions are retained for audit, and the system generates its own compliance record. No one would consider deploying a new rating algorithm by editing a configuration file with no version control. That standard of infrastructure is exactly what is needed for AI agents.

When agent behavior is embedded in code, compliance teams cannot access it without engineering support. Storing agent definitions as versioned configuration records changes this - any authorized reviewer can see exactly what instructions any agent version was operating under at any moment in time. Built on that foundation, the framework can enforce the lifecycle mechanically, log every decision with the version that produced it, surface override patterns automatically, and generate regulatory evidence on demand.

The carriers that have built this treat governance as an engineering problem, not a policy exercise.

Where to Start

Start with the agent registry. For each AI system in production, name it, document what it does, record the live version, and identify its owner. Add prompt version control before the next agent change, and decision logging before the next production deployment. Governance built incrementally as infrastructure - before the examination, before the finding, before the failure - is faster to production and more durable than governance assembled in response to one.

The regulatory direction is clear, and the examination questions are already being asked. The difference between carriers that answer them confidently and those that cannot is infrastructure that existed before the question arrived.

Has a family member ever given you a gift you can't bear, yet can't refuse, and it simply becomes part of the decor? It might be a decanter so impractical that it's ornamental, but it has to be brought out every time they come over; or a portrait that asks fundamental questions about the nature of your relationship, yet over time you no longer register that it's there. 

Operational challenges can be like this; unwanted gifts that become clutter, obstacles that are easy to rationalize. As they accumulate, they require incremental effort to navigate and leach efficiency. Yet when we approach a familiar operational challenge from inside the organization, we risk framing the challenge so narrowly that we're boxed in with too few options available. We refer to this as approaching challenges through a lens of our experience - and it can become part of the problem, rather than a means of solving for it.

Seeing a problem through the lens of our experience describes a way of seeing that includes all our knowledge of the history of the problem. All the attempts to resolve it, the failures, the frustrations; it's the voice that says, "We've tried that before and it didn't work." Returning to our furniture metaphor, it's not dissimilar to saying, "We can't move that painting. We took it down once, and my brother got upset." The lens of experience is effective at keeping you on the same track but it's less likely to help change direction. Evaluating a persistent operational challenge through a lens of expertise is vastly more effective.

Approaching a familiar challenge through a lens of expertise means stepping outside of the challenge, viewing it more objectively, and applying our knowledge to that problem. This is the secret sauce of consulting, the classic "outside-in perspective," yet it's possible to strengthen this capability within your own organization. The key is understanding how changing the structure of a problem helps to create new ways of seeing it. By carefully evaluating a problem and adjusting its constraints, experienced operators can see a familiar challenge with a broader perspective, and then bring their hard-won expertise to bear.

I worked with an insurance property repair firm whose leaders shifted their focus from a lens of experience to a lens of expertise with spectacular results. They were part of an insurer's repair vendor panel and found themselves competing across a broad range of repair categories, tackling jobs that ranged from minor fence repairs and garage doors through to major reconstructive work for insureds. Smaller jobs only required general handymen - low cost, low risk, and the pool of available contractors was broad - whereas the larger jobs required more skilled trades and more oversight - higher cost, higher risk, and a narrower pool of trades. Larger firms on the panel could absorb the occasional job that went off the rails, but this firm was small enough that even one or two jobs that went over budget hit profits hard. That was the model. Until this firm opted to re-imagine and renegotiate their panel membership.

The repair firm reimagined their business in two stages: first, they negotiated with the carrier to remain on the panel as a "small repairer." They would only accept smaller repair work but take higher volumes. This was feasible because the pool of trades was large and - given the nature of largely weather-related property damage - jobs were often geographically co-located. One trade could attend multiple sites in a day, which allowed for bundling and improved efficiency. In exchange, the repairer would offer a reduced rate because they weren't subsidizing larger jobs.

Second, they re-designed their operations from within by re-structuring their project management approach. They turned the entire model upside-down, from how they hired trades and retained them to how they would project manage each job. Each repair was broken into its discrete segments (plastering, painting, electrical, and so on) and were arranged such that the right trade attended at the right time - a virtual production line. Trades tapped in and out on their cellphone app, which gave the business visibility of their activity, plus allowed for them to estimate the time required for each job - a feedback loop that informed project, pricing, and contract-hiring forecasts.

The results were significant. The carrier ultimately integrated the model directly into its property claims flow, allowing customers to move from first notice of loss to completed repairs with a speed that hadn't previously been possible. Customer satisfaction ratings exceeded 90%. The firm had transformed itself not by responding to competitive pressure, but by isolating the fundamental conditions of their business and restructuring them to reveal entirely new ways of operating.

Your operations function may be more or less complicated than this example, but there's likely at least a handful of persistent challenges you'd love to unpick. Start by examining the assumptions and constraints that shape how you interpret the problem. Change those, and new solutions will follow.

Artificial intelligence is already changing insurance claims operations. It can shorten cycle times, improve fraud detection, reduce administrative costs, and help carriers handle routine claims with greater speed and consistency. Those benefits are real. But the difference between a useful AI system and a risky one is rarely the model itself. It is the control environment around it. 

After 15 years in financial operations across telecommunications, banking, and healthcare, I have learned that systems do not usually fail because they produce outputs. They fail because organizations do not build the right controls for what happens when those outputs are wrong. That lesson is especially relevant in insurance claims, where AI can recommend payments, trigger denials, or escalate fraud investigations at speed and scale.

This is why the most important question in claims AI is not whether a model performs well on average. It is what happens when it does not. Who reviews the outlier decision? What happens when source data is incomplete or inconsistent? Which claims are allowed to move straight through, and which require human judgment? Without clear answers to those questions, automation creates exposure faster than it creates value. 

The insurance industry has made real progress. Many carriers now use AI in some part of claims handling, especially for low-complexity workflows. But mature deployment remains limited. The gap is not just technical. It is operational. Insurers often struggle with fragmented data, inconsistent workflows, weak escalation paths, and governance models that are more aspirational than enforceable. In practice, that means claims AI often performs inside silos rather than inside a coherent control framework. 

In financial operations, this kind of weakness is familiar. I have seen organizations lose significant revenue not because the systems were incapable, but because no one had defined what should happen when an exception appeared. In one credit control role, I identified more than $10 million in revenue leakages. Those leakages persisted not because no system existed, but because process gaps allowed errors to go unchallenged. Claims AI creates the same risk, except with higher speed, broader scale, and greater regulatory sensitivity.

So what guardrails actually work?

Human review for non-routine claims. Straight-through processing can be appropriate for low-value, low-complexity claims where the decision logic is narrow and well tested. But once a claim involves material exposure, medical complexity, ambiguity in coverage, or fraud indicators, human judgment must re-enter the process. This is not resistance to AI. It is sound risk design.

Explainability for adverse decisions. If an AI system recommends denial, escalation, or fraud review, the rationale must be understandable to the people accountable for that outcome. An adjuster cannot meaningfully supervise a recommendation that cannot be explained in plain terms. Explainability is not just a technical preference. It is the basis for accountability, defensibility, and fair review.

Continuous data-quality control. AI systems do not fail only because of bad models. They also fail because of incomplete, stale, fragmented, or poorly governed data. In claims operations, a data issue is not a minor defect. At scale, it becomes a multiplier of bad decisions. Regular review of upstream data sources, transfer points, and exception patterns is essential.

Defined exception and escalation pathways. Every model has edge cases. Effective governance assumes this from the start. Claims that fall outside confidence thresholds, conflict with policy logic, or present unusual fact patterns should move automatically into a structured review queue with identified owners and documented next steps. In strong operating environments, exceptions are not left hanging. They are routed.

Active regulatory monitoring. AI governance in insurance is no longer an internal policy matter alone. Carriers now operate in an environment of increasing scrutiny around disclosure, fairness, bias, consumer protection, and human oversight. Any organization deploying AI in claims must treat compliance monitoring as part of the operating model, not as an afterthought.

It is equally important to be clear about what does not work.

Principles without enforcement do not work. A statement about responsible AI is not a control unless it is backed by auditability, accountability, and operating discipline.

Black-box decision making in high-stakes contexts does not work. A model that cannot be explained may still produce accurate outputs in aggregate, but it creates real risk when applied to adverse decisions that affect claimants and attract scrutiny.

Deployment on unvalidated source data does not work. AI does not fix weak data foundations. It accelerates the consequences of them.

Minimal staff training does not work. Claims professionals do not need to become data scientists, but they do need enough AI literacy to interpret outputs, question recommendations, recognize limitations, and escalate when needed.

The operational stakes are high. Carriers that deploy AI well can improve speed, consistency, and cost performance. Carriers that deploy it poorly can create regulatory exposure, claimant harm, and reputational damage that overwhelms any efficiency gain.

In the end, the real issue is not whether AI belongs in claims. It does. The issue is whether insurers will build the operational discipline required to make AI trustworthy. The winning organizations will not be the ones with the most impressive demos. They will be the ones with the clearest controls, the strongest escalation design, the cleanest data discipline, and the most accountable governance.

AI can make claims operations faster. Only guardrails make them reliable.

Every major insurer has an AI strategy. Most of them will fail. Not because the technology isn't ready — it is. Not because the use cases don't exist — they do. The strategies will fail because organizations treat AI as a point solution rather than a platform, and they underestimate how fundamentally it demands a different operating model.

I spoke about this at ONUG (Open Networking User Group) last fall under the title "Beating the 4%: Why AI Fails." The thesis is straightforward: the vast majority of enterprise AI initiatives stall at the pilot stage, not because of technical limitations, but because of misalignment between technology investments and business operating models. Insurance, with its complex distribution relationships and legacy infrastructure, is particularly exposed to this failure mode. The question for carriers isn't whether to adopt AI. It's whether they have the architecture — organizational and technical — to make it stick.

Building the Right Foundation

The organizations that succeed with AI don't start with models; they start with alignment. Before any model goes into production, the business objective has to be clear, the data must be trustworthy, and the teams have to understand what the tool is solving and why. That sequencing matters more than the technology itself.

A consistent pattern across successful AI programs is centralized governance, a single framework through which development and deployment are coordinated. This isn't bureaucracy. It's how you prevent fragmentation. Without it, you get dozens of disconnected pilots, inconsistent data practices, and no coherent path to scale. With it, you build institutional muscle: teams that know how to evaluate, deploy, and improve AI solutions within a common framework.

Early applications often show up in operational improvements – streamlining workflows or improving access to knowledge — but those are table stakes. They're the foundation, not the destination.

Owning the Experience Layer

The real competitive battle in insurance isn't being fought in the back office. It's being fought in the experience layer — the digital surface where financial professionals and clients actually interact with your products and your brand. Carriers that own that layer will win. Those that cede it to distributors, aggregators, or fintechs will spend the next decade competing on price alone.

The real opportunity lies here: embedding AI directly into the workflows of financial professionals, not as a separate tool they have to context-switch into, but as intelligence woven into the systems they already use. That can take many forms – from meeting preparation and product insights to tools that help advisors refine how they engage with clients and improve over time.

The unifying thread is data. Personalized, trustworthy, AI-powered experiences are only possible when data is unified enterprise-wide. Without that foundation, you are not personalizing — you are guessing.

Democratizing AI Across the Organization

I've seen this pattern play out across industries – insurance, real estate, professional sports, and media & entertainment. The organizations that win with transformational technology are never the ones that centralize it in an IT function and call it done. They are the ones that democratize it: making AI capability accessible, legible, and useful to colleagues across every function.

The organizations making real progress are not just building AI tools – they are building AI fluency: helping teams understand how to interpret outputs, where to trust automation, and where human judgement remains essential.

The Carriers That Will Lead

With $483 trillion in projected retirement savings shortfalls by 2050, the demand for trusted financial guidance is only going to intensify. The carriers that are positioned to meet it will not be the ones with the most AI tools. They will be the ones that have built the right operating model, owned the experience layer, and treated AI not as a pilot project but as organizational infrastructure.

That is a harder problem than it looks. But it is the right one to solve.