Litigation is one of the most volatile and least standardized components of indemnity leakage. Venue outcomes vary widely, plaintiff bar strategy has grown more sophisticated, and jury behavior remains inherently unpredictable. This is exactly why litigation is often the hardest leakage driver for claims organizations to bring under control in day-to-day operations.

According to a 2026 CLM Litigation Management Study, auto and trucking has overtaken general liability as carriers' single largest litigation spending driver. It now accounts for 33% of total litigation spending, up from 24% in 2023. At the same time, data from Marathon Strategies' May 2025 analysis puts the median nuclear verdict against corporate defendants at $51 million, up from a $44 million median. Even mid-severity claims now carry thermonuclear tail risk in adverse venues.

Claims organizations at P&C insurers are stretched thin. However, when done well, claims litigation management is one of the most effective methods for controlling severity and protecting loss ratio performance.

A Three-Layer Approach to Claims Litigation Management

The CLM 2026 data also show that settlement — not trial — remains the dominant path to resolution. A mean of 87% of non-workers' comp litigated claims settle, against a verdict rate of just 2.9%, indicating that settlement quality, not just avoiding trial, is the true performance lever.

Yet many legacy claims processes were not built for the circumstances insurers face today. Litigation propensity is assessed too late to change outcomes. Panel counsel is retained based on legacy relationships rather than performance. And even once a case is flagged, panel counsel selection and case strategy tend to move independently of one another, with reserves catching up only after the fact.

Closing that gap requires more than incremental fixes. It calls for an end-to-end approach built on three integrated layers, each tied to a distinct point of loss ratio impact.

1. Analytics and AI

The foundation of a modern litigation management approach is intelligence applied at the earliest possible point in the claims life cycle. Machine learning models embedded at first notice of loss (FNOL) score litigation likelihood using injury type, claimant history, adjuster notes, and jurisdiction, giving claims teams a head start on cases most likely to become contested. The earlier that signal arrives, the more severity is still avoidable.

Attorney selection benefits from that same discipline. Panel counsel are scored on cost per claim, resolution time, and win rate, then matched to cases based on demonstrated performance rather than tenure or familiarity. Later in a case's life cycle, predictive models take on a different task, comparing the likely cost and outcome of settling now against continuing to litigate so adjusters can weigh that decision with data instead of instinct.

A Litigation Risk Index ties these capabilities together: a composite score built from plaintiff attorney specialization, reptile-theory signals, and jurisdiction severity. In practice, the score is used to trigger escalation thresholds, reserve reviews, and counsel reassignment, giving claims organizations a consistent way to gauge venue risk across an entire portfolio.

2. Claims Management Platform

Where analytics identify risk, the claims management platform puts that intelligence to work inside the daily workflow, rather than leaving it in a separate tool an adjuster has to remember to check.

Preferred counsel assignment is automated first. Panel tiers are enforced by default, and any deviation is flagged for supervisor approval so that attorney selection is strategic, not based on firm availability or convenience. Each file carries a real-time litigation risk score inside the platform, visible to the adjuster the moment a case is assigned.

From there, oversight becomes continuous instead of periodic: a centralized case tracking dashboard, automated diary management, and billing review keep reserve alignment current as a file moves through each stage. The Litigation Risk Index recalculates at key case events, automatically escalating a file to the litigation management track the moment its risk profile changes. The result is fewer files drifting off-process, and less variance in how similar cases get handled.

3. Advisory and Governance

Analytics and platform automation are only as effective as the human judgment guiding them, which is where advisory and governance close the loop. Quarterly venue heat map updates, paired with judicial behavior analysis (plaintiff-friendly tendencies, historical award patterns, and jurisdiction-specific trends) give litigation leadership an continuing view of exposure and reserve guidance that shifts as jurisdictions evolve, rather than a static assessment revisited only after a bad outcome.

That intelligence feeds directly into panel strategy. Rather than reviewing counsel performance on an ad hoc basis, this layer calls for a structured panel redesign process: identifying underperforming firms, restructuring fee arrangements, and setting outcome targets tied to results. A regular litigation strategy review cadence carries that discipline into individual files, with adjusters and defense counsel working from file-level guidance and early resolution playbooks, particularly for cases trending toward nuclear venues, well before a verdict is at risk.

This layer also defines what happens when risk scores change. Intervention protocols tied to low, medium, and high score bands establish clear escalation paths, agreed upon with carrier leadership in advance, so no file's rising risk profile goes unaddressed for lack of an owner.

When these three layers operate together rather than in isolation, litigation shifts from a cost absorbed after the fact into a severity lever that is actively managed from FNOL forward.

Turn Litigation Insight Into Action

Litigation is not going away, and neither is the pressure it places on loss ratios. Exposure risk will likely continue climbing, and nuclear verdicts show little sign of slowing down.

No approach eliminates litigation risk entirely. But carriers that manage it proactively through integrated analytics, embedded workflows, and disciplined governance can meaningfully influence the trajectory of a case rather than simply reacting to it after exposure has already escalated.

To learn more about navigating the changing environment around litigation management, read ValueMomentum's whitepaper "Combating Social Inflation: Strategies for Claims Organizations to Reduce Leakage."

Generic AI is reaching its limit in insurance. It can summarize a submission, draft a customer response or extract information from a claims file. What it cannot do independently is make the high-stakes decisions that determine insurance performance.

Those decisions require more than a plausible answer. They depend on an insurer's data, pricing and underwriting rules, risk appetite, regulatory obligations, portfolio position and operational context. Without that grounding, AI may produce an answer that sounds credible but cannot be trusted, explained or acted upon.

According to Grant Thornton's 2026 AI Impact Survey, 44% of insurance executives said governance or compliance challenges had caused AI projects to fail or underperform, while 56% identified regulatory uncertainty as a leading barrier to scaling AI.

This is driving the industry toward vertical AI built specifically for insurance. Unlike general-purpose tools, vertical AI can apply intelligence within the context, controls and accountability that real insurance decisions demand.

The Danger of Faster Silos

Most insurers already have vast amounts of data, but rarely a complete view at the moment a decision must be made. Policy data may sit in one system, claims history in another and customer interactions somewhere else. Portfolio information and external risk signals may also arrive on different cycles.

Each source tells part of the story, but none necessarily reflects the full business context.

When insurers place AI on top of that partial view, they may produce an answer more quickly without producing a better decision. In high-stakes decision making, speed cannot compensate for missing context.

Insurance has traditionally been organized around separate disciplines, including pricing, underwriting, claims, customer engagement and portfolio management, even though the outcomes of those functions are closely connected. A pricing decision affects conversion and retention. An underwriting decision changes portfolio quality. A claims experience can influence renewal, customer value and future risk.

The business understands these relationships, but its systems do not always reflect them.

The risk is that insurers layer AI onto the same fragmented operating model. Pricing, claims, underwriting and service may each gain tools, while the organization still lacks a shared understanding of the decision being made. Individual tasks become faster, but the underlying disconnect remains.

That is the limit of horizontal AI. It can improve an activity without improving the enterprise decision surrounding it.

Insurance AI Should Follow the Decision

Policy administration, billing, claims and CRM platforms remain essential to insurance operations. They preserve records, process transactions and provide the stability insurers rely on. AI should strengthen the work performed within and across those systems, rather than attempt to replace them.

These platforms are highly effective at recording what has happened. The greater opportunity lies in helping insurers determine what should happen next.

A renewal decision, for example, may require policy history, recent claims, customer value, portfolio concentration, emerging risk signals, a pricing model, underwriting appetite and current regulatory requirements. No single platform owns all of that context.

The architecture must therefore be decision-centered rather than core-centered. Insurers need to identify the intelligence, rules, workflows and human expertise that must come together around each decision, then orchestrate those elements across the systems they already use.

Vertical AI extends far beyond adding insurance terminology to a general-purpose model. It is designed around how insurance work is performed, how decisions are governed and how their consequences are managed.

A rate change, for example, affects more than the price presented to a customer. It can influence portfolio composition, retention, profitability, fairness and regulatory filings. The AI supporting that decision must understand those connections.

Bringing the necessary context into the workflow may require predictive AI for risk and pricing, generative AI for explanation and guidance, and agentic AI for bounded, multistep execution. The type of model matters less than the insurer's ability to orchestrate the right intelligence around the decision.

The decision must come first.

Governance Must Travel With The Action

Insurance has always delegated authority within defined boundaries. Controls throughout the process allow insurers to distribute decision-making without losing accountability.

Agentic AI changes the risk profile because it can take action rather than simply provide advice. When an AI system can initiate or execute steps in real time, reviewing its activity after the fact may be too late.

Governance must therefore be embedded within the decision itself. It should define which data can be used, which models and rules may be applied, when human review is required and how the complete decision path can be reconstructed.

Not every AI component will be deterministic, but the route from input to action must be controlled wherever the decision demands it. Probabilistic intelligence requires deterministic guardrails, including defined permissions, monitoring, escalation paths, and the ability for a person to intervene.

The regulatory direction is already clear. The NAIC's model bulletin calls on insurers to establish governance, documentation, testing, validation and third-party oversight for AI systems supporting decisions that affect consumers. It also reinforces that insurers remain accountable for those decisions, regardless of the technology used to support them.

Human oversight must be designed into the operating model. This does not mean asking someone to rubber-stamp every AI-generated output. It means defining what can be automated, what must be reviewed, when escalation is required, and who ultimately carries responsibility.

Done well, this approach also gives experienced underwriters, actuaries and claims professionals greater leverage. Their hard-won expertise can become reusable decision logic, business rules and escalation paths, rather than remaining trapped in documents, manual processes or the knowledge of a small number of individuals.

Measure Decisions, Not Deployments

The insurance industry has invested heavily in data, analytics, predictive models, cloud platforms and core modernization. The remaining gap lies between those assets and the decisions that run the business.

A vertical approach allows insurers to advance AI without waiting for a multiyear core replacement. It can draw on the systems, data, and models insurers already trust, then orchestrate them around the decisions that need to improve today.

This makes existing investments more valuable while giving the business greater flexibility to adapt, without destabilizing the systems that keep operations running.

Success should therefore be measured by the quality and business impact of the decisions being improved, not by the number of AI tools deployed. That may include stronger pricing performance, more consistent underwriting, faster product changes, improved customer outcomes, or better portfolio management.

Insurance is an industry of connected consequences. AI built for isolated tasks will miss those connections, regardless of how capable the underlying model may be.

Insurance-native, vertical AI begins with the decision. It brings together domain context, predictive and generative intelligence, business rules, workflows, governance, and human judgment at the point where action occurs.

That is how insurers can move beyond isolated AI use cases and turn AI into a trusted, scalable capability for the decisions that matter most.

Have you ever calculated the true cost of a manual policy update?

Mid-term policy modifications carry heavy administrative overhead across insurance operations.

• Adding a vehicle to a commercial fleet requires significant data entry.
• Adjusting coverage limits involves pulling up multiple internal systems.
• Updating a mortgagee address demands careful attention to detail.

These routine tasks pull account managers away from meaningful client interactions.

High-touch service relies on speed, accuracy, and clear communication. Automating endorsement processing preserves human energy for relationship management. Account managers were hired for their expertise in risk management and their ability to build client trust. Every hour spent manually keying data is an hour taken away from proactive client service. Insurance leaders should look at back-office workflow automation as a primary strategy for protecting client relationships.

Operational Friction Points in Endorsement Management

Endorsement requests often arrive through unstructured emails containing the insured's instructions. Account managers spend their valuable time manually extracting information and re-keying data into the AMS and carrier portals. This repetitive administrative overhead creates operational bottlenecks.

When routine endorsements take days to process due to back-office backlogs, client trust begins to erode. Insureds expect quick turnarounds for mid-term changes. Delayed processing affects the client relationship right when individuals and business owners need confirmation that their assets are protected.

What Intelligent Endorsement Automation Looks Like

Agencies can modernize these workflows without losing operational control. Purpose-built insurance automation handles repetitive data entry while keeping licensed staff in control of key decisions.

Request Intake and Data Extraction

Intelligent tools can continuously monitor shared email inboxes and carrier portals for incoming endorsement notifications and direct change requests. Using advanced data extraction techniques, these systems read unstructured emails and attached PDFs. The software automatically identifies key field data, pulling policy numbers, effective dates, coverage adjustments, and premium changes directly from the source material.

Policy Validation and System Matching

Extracting data requires accuracy before any record changes take place. Automation cross-references extracted request data against active policy records in your core AMS. The system verifies endorsement details, checks for missing information early in the workflow, and confirms that requested coverage limits or deductibles fall within allowed ranges. Early validation eliminates time-consuming back-and-forth communication with clients.

Automated System Updating

Once validated, automated tools apply verified changes directly to the correct policy record in the AMS. Automation updates coverage limits, adjusts premium amounts, and modifies effective dates without requiring manual data entry from staff. Detailed audit trails capture every modification automatically, logging timestamps and specific changes to maintain complete compliance.

Exception Handling and Human Oversight

Automation functions best alongside human judgment. When a complex request comes in, such as a major shift in coverage boundaries or missing documentation, a good insurance automation solution flags the file for manual review. An assigned account manager receives an immediate notification to evaluate the exception, ensuring complete oversight over non-standard adjustments.

Turning Back-Office Efficiency into Better Client Experience

Streamlining back-office workflows through automation translates directly into a better client experience. Shifting staff focus away from manual data entry reclaims capacity for high-value client engagement. Account managers gain the time needed to review coverage needs proactively, discuss risk management strategies, and prepare for upcoming renewals in advance.

Automated extraction and validation also eliminate human error. Re-keying information manually can lead to typos, which sometimes cause coverage gaps or billing discrepancies down the line. Consistent logic across workflows reduces variance and ensures records remain accurate across all systems.

With endorsement processing automation, clients receive quick confirmations and updated policy documents. Rapid service builds confidence. Processing policy change requests swiftly and accurately reinforces the client's confidence in your agency's capabilities.

Elevating the Client Experience Through Automation

Insurance agencies should view automation as an enhancement to human potential. By removing repetitive administrative tasks, account managers can return to the relationship-focused work that drives client retention. Modern solutions layer directly onto existing agency management systems and carrier portals, driving efficiency without disrupting core operations.

The insurance industry has spent the summer treating storm count as a proxy for risk. With NOAA's 2026 Atlantic outlook pointing to a below-normal season, the prevailing read across multifamily ownership has been one of relief after several punishing years.

But that read conflates two things that are not, in fact, the same variable.

The industry's own reinsurance modeling doesn't support the equivalence, and neither does the loss pattern showing up in ordinary, non-catastrophic claims that have nothing to do with named storms at all.

Start with what a "quiet" season actually means statistically rather than what it implies emotionally. A below-normal story count reduces the probability of a major landfall, but it does nothing to reduce the severity of one if it happens. And, it also says nothing at all about the far more frequent, far less dramatic losses, from fires and burst pipes to structural failures, which drive claims in any given month regardless of what the Atlantic is doing.

That second category is where the coverage gap actually lives, and it's been on display twice recently. In Henrico County, Va., a fire at the Darby House senior apartments displaced dozens of residents, prompting the Virginia Apartment Management Association to remind tenants publicly that a landlord's policy stops at the building's walls; it doesn't cover what's inside a resident's unit. Days later, a roof collapse at North Union Apartments in Midvale, Utah, left residents uncertain whether their losses were covered at all, with coverage varying from resident to resident depending on who had signed up and who hadn't.

Neither event involved a storm, and that's precisely why they're instructive. They show the coverage gap operating at full strength in the absence of the peril everyone's currently watching.

If a below-normal hurricane season lulls owners into treating resident-level insurance enforcement as a lower priority this year, the Henrico and Midvale losses are what that complacency looks like even in a quiet season's ordinary months, before a single storm enters the equation.

"Quiet" is a statistical average, not a guarantee

NOAA's outlook has pointed toward below-normal activity all summer, with only four named storms recorded so far. Colorado State's forecasting team, meanwhile, has revised its numbers downward since spring.

On paper, this is good news for a market that's spent the last several years absorbing catastrophic losses. Swiss Re's own data complicates that reading, however.

The reinsurer's institute found that 2025's global insured catastrophe losses of $107 billion came in below the long-term trend almost entirely because no major U.S. hurricane made landfall that year. This was characterized explicitly as favorable variability, not a reduction in underlying risk.

If 2026 tracks the long-run 5-7% annual growth trend that's held since 1996, the same report projects insured losses climbing to roughly $148 billion; in a peak-loss scenario driven by a single major hurricane or earthquake, that figure rises to $320 billion.

That's the crucial detail a headline storm count doesn't capture. The industry's own reinsurance modeling treats "quiet so far" as noise sitting on top of a rising baseline, not as a trend in its own right.

Coverage of this year's season has made the same point more bluntly, noting that a below-normal forecast produced the exact conditions when Hurricane Andrew struck in 1992, remembered as one of the costliest storms in U.S. history. The industry as a whole must be reminded, therefore, that storm count and loss severity aren't the same variable. They never have been.

For multifamily owners, the practical takeaway isn't that this season will necessarily be bad. It's that a favorable forecast is not a reason to relax the parts of the risk picture within an owner's control. Because the parts that aren't, including where a storm tracks and how intense it becomes on approach, remain exactly as unpredictable.

The gap sits below master policy

While the macro conversation stays fixed on named storms, a far more mundane and fixable problem recurs every month. Most residents simply aren't required to carry coverage for the fire, burst pipe, and structural failure losses that actually drive claims, the kind that hit Henrico and Midvale last month.

A CNBC Select analysis of a RentRedi study, for one, found that only about 47% of U.S. landlords required renters insurance in 2025, leaving a majority of leases nationwide with coverage effectively optional, at an average cost of $14.25 a month, according to the Insurance Information Institute.

This is a strikingly low enforcement rate for a policy this inexpensive, and this directly tied to an owner's own liability exposure.

That gap also compounds a cost problem operators are already living with on the master-policy side. The NAA's Premium Pulse analysis, drawing on Federal Reserve data, found that real, inflation-adjusted multifamily property insurance costs per unit rose about $465 in 2019 to $821 in 2024, representing a 77% increase in five years. Insurance, in fact, now represents 4.8% of multifamily revenue nationally, up from under 2% in 2000.

The Federal Reserve Bank of Minneapolis reached a similar conclusion when surveying operators directly, documenting firsthand accounts of premiums rising sharply even as coverage itself shrinks.

Owners are absorbing that cost increase on the building side while leaving an inexpensive, liability-reducing tool underused on the resident side, and that mismatch gets harder to justify the more premiums climb.

Some operators have tried to solve this the other way, by folding tenant-caused damage into a landlord-side Tenant Legal Liability program instead of relying on individual renters' policies. Subrogation counsel, however, has flagged a structural cost to that shortcut.

Because TLL coverage is written to protect the landlord's own insured interest rather than the tenant's, it can limit or bar the landlord's insurer from pursuing subrogation against a negligent tenant once a claim is paid.

A program built to simplify risk management can therefore end up narrowing the very recovery rights that keep loss from landing entirely on the portfolio. Individual, verified renters insurance, carried at adequate liability limits rather than the token minimums many leases still specify, preserves that recovery path in a way blanket landlord-side products generally don't.

Waiting to see the season

The fix here isn't a data availability problem in the way it might sound. Running an effective renters insurance program requires knowing who's covered, at what limits, verified against which unit, and that information exists in principle at every property.

What's missing in most portfolios is a consistent, centralized way to confirm it at scale and catch the gap before a claim exposes it, rather than after. Enforcement, in other words, isn't a question of building new infrastructure from scratch, but one of actually verifying what leases already claim to require, instead of accepting a self-attestation and moving on.

That's a lever owners control entirely on their own timeline, independent of carrier appetite, reinsurance pricing, or how the rest of this hurricane season unfolds.

If the season stays quiet, tighter enforcement still reduces claims exposure and preserves subrogation rights against the ordinary losses that happen in any given month, storm or not.

On the other hand, if the update doesn't hold, the operators who tightened enforcement now won't be the ones relearning this lesson in October.

When two steel columns buckled on the 21st floor of the former Pfizer headquarters in Manhattan this July, the building safety industry watched a familiar sequence unfold, and commercial property insurers should have watched just as closely. 

Reporting noted that engineers were cautioned against assuming the visible damage marked the full extent of the problem at the 37-story tower, then mid-conversion into more than 1,600 apartments. Structural elements are connected, and a failure in one area often signals stress elsewhere. Early accounts pointed toward a missing steel reinforcement plate, a detail that should have appeared in the project's engineering plans but was never installed, according to engineers cited by The Real Deal.

For a carrier underwriting that asset, or an adjuster assigned to the claim, it would be easy to call this a documentation failure. That framing undersells the problem. The tower did not lose its documentation. It lost records continuity. Every renovation, code cycle, and material change was tied to whatever system or individual happened to be tracking it at the time. When those links broke, so did the file's ability to describe the true risk on the books.

The missing layer is not another document repository. It is a persistent identity connecting every record, inspection, permit, renovation, owner, engineer, and platform to the same physical asset across its lifecycle and every policy period it will ever carry.

Documentation Answers the Wrong Question

Even carriers with disciplined underwriting files run into this. A document repository answers what records exist. It cannot answer what actually happened to a specific asset across its operating life, the question a claims examiner is really asking.

A recent industry analysis of facility continuity found that as veteran facility managers retire, decades of undocumented building knowledge often leave with them, since it lives in someone's head rather than in a system. The analysis cited IFMA projections that more than 45% of facility management professionals worldwide will retire within a decade, removing much of the institutional memory insurers have quietly relied on when pricing risk.

That is the same failure described two ways: a retiring manager who carried undocumented knowledge out the door, or a file that cannot say what a structure can support after decades of alterations. The cause is identical. The record lives with the software, the vendor, or the employee, never with the infrastructure itself.

What Claims and Underwriting Actually Require

In a post-failure scenario, structural engineers, investigators, and claims teams are not simply looking for a folder of drawings. They need a reliable, continuous chain of custody for every material change, traceable to who made it, when, and under what code cycle. A binder handed off at turnover cannot provide what's needed.

The same need extends well beyond a single claim. Fire departments, FEMA, adjusters, building officials, and search and rescue teams all depend on the same continuous history of what a building is, what it can support, and what has changed. So does every actuary modeling portfolio exposure, and every reinsurer pricing a treaty against it.

What claims and underwriting actually require is closer to a VIN for physical infrastructure, a persistent identifier every contractor, engineer, permitting office, carrier, and platform can reference over decades, regardless of who owns the data or which system created it. Imagine underwriting an aircraft fleet if every maintenance record moved to a new numbering system each time ownership changed. That is unacceptable in aviation. It remains normal for buildings.

A Survey That Is Measuring Something Deeper

Viewed this way, recent survey findings from ARC Facilities on facility leader confidence are not really measuring documentation quality. They are measuring identity fragmentation, the same fragmentation showing up as inconsistent exposure data across a book of business. The survey found many facility leaders describe only partial confidence in accessing critical building information when it matters most, one noting their organization lacks a reliable system.

Low confidence in complete records is what a carrier's risk engineers should expect in an industry where every renovation, ownership change, and software migration creates another chance for the thread between an asset and its history to snap, often just before a claim tests it.

The Layer the Industry Never Standardized

Over recent decades, the built environment has standardized layer after layer of how infrastructure gets built and managed. CAD standardized design. BIM standardized information. GIS standardized location. IFC standardized interoperability. Digital twins standardized representation.

None of those layers standardized identity, a persistent, portable reference that follows infrastructure across every system, every underwriting cycle, and every claim, throughout its lifecycle. Each platform generation has improved how information about an asset is captured, but none has solved for what happens once the platform, vendor, or risk manager who understood it moves on.

A digital twin cannot remain continuous if the identity of the asset it represents is not continuous. Neither can a book of business.

Interoperability tells systems how to exchange information. Identity tells systems what they are exchanging information about. The industry has spent 50 years perfecting the first problem while leaving the second unaddressed, and insurers have been pricing risk on top of that gap.

A Different Question for Risk and Insurance Leaders

The Pfizer conversion is useful not because documentation failed at a single point, but because continuity failed across an entire history of changes, and continuity depends on identity. That is a claims story and an underwriting story at once.

The more productive question for the insurance and risk industry isn't how thoroughly infrastructure gets documented. It's why so little of it carries a persistent identity across ownership changes, renovations, and software migrations, including changes in carrier, broker, and policy. Every industry exchanges data about buildings without a shared way to identify what it describes.

The built world has standardized nearly every way information is created, exchanged, and analyzed. The next standard will not be another data format. It will be persistent identity, and the carriers who adopt it first will be the ones who can actually price the risk they are holding.

In 2012, someone showed me a life insurance policy still in force. Issued in 1928. I'd come to the sector after a long run in credit cards, where you know what you're carrying within six to 18 months. I did the math without meaning to: that policy predated Social Security.

Insurance carries risk differently, and the difference isn't just duration — it's what either side can do about it once the ink is dry. A lender can act on an individual account: reduce a line, renegotiate, price it differently, charge it off. A life insurer cannot unilaterally touch a policy already in force; that right belongs to the policyholder, not the carrier. The industry doesn't just carry decades-long risk. It carries it locked, one contract at a time, with no lever to revisit terms even after the assumptions underneath them have quietly stopped matching reality.

That's an issue the industry's own machinery was never built to answer. Reserving, asset adequacy testing, hedging, reinsurance, the sale of legacy blocks to run-off specialists — all of it real, all of it sophisticated, all of it built to answer whether the aggregate is still solvent. None of it answers a narrower, harder question: does this one assumption, buried in a block that's still passing in aggregate, still hold? The standards touching assumption review ask for judgment, triggered by whatever surfaces the need — not a date on a calendar. The tests that do run on a schedule check the aggregate, not the assumption underneath it.

Here's the tell that this gap stopped being theoretical

In March 2026, the NAIC began piloting a structured AI Systems Evaluation Tool across 12 states. It isn't a general AI-governance gesture. Examiners using it can ask insurers to detail the oversight of AI systems used by actuarial service providers specifically, and to walk through the testing, verification, frequency, and methodology behind them. For actuaries who build or validate predictive models — which increasingly means experience studies and assumption-setting, not just pricing or claims — that means model documentation and assumption justifications becoming subject to regulatory examination, not just internal actuarial standards of practice. That's not an analogy to a proactive posture. It's regulators reaching directly into how AI-touched assumptions get tested and justified — a first, concrete step toward exactly the granular scrutiny this piece is arguing the industry doesn't otherwise have.

(A brief aside: the industry has pushed back hard on this pilot, and not merely over workload. Trade groups' formal comments call it "voluntary for regulators while compulsory for companies," note it has no defined end date, and flag that findings from what's explicitly an exploratory phase could be held against a carrier before the tool itself is finalized. Those are fair governance objections. They're also, in a way, evidence for this piece's own argument: regulators are building this proactive posture iteratively, through judgment and a pilot, before hardening it into a fixed rule — which is closer to the muscle insurers themselves need than a reason to dismiss it.)

That same year — 2012 — I watched a version of this exposure play out in real time, from inside a different seat, in an industry-wide effort that was already well underway. The 2008 crisis had left insurers holding large in-force books of annuities carrying guaranteed crediting or withdrawal rates set in a much higher-rate era. When rates fell and stayed down for years, those guarantees didn't just look generous — they became a standing liability the original pricing had never been stress-tested against. And because nobody can unilaterally rewrite a policy already in force, the fix was never going to be a model update. It was years of patient, unglamorous, one-by-one work: building offers attractive enough that policyholders would voluntarily give up guarantees the industry could no longer comfortably afford to keep. Carrier by carrier, contract by contract, across the sector — not because anyone lacked talent, but because nobody had a standing practice of asking, before the rate environment turned, whether that guarantee still made sense.

That's the exposure insurance carries now, in its own way, with its own kind of contract. It isn't new, and it was never really about AI — the gap between "the aggregate is fine" and "the individual assumption still holds" predates any model running against an underwriting file. What's changing is the cost of not knowing. Checking assumptions individually, continuously, used to be a resourcing decision nobody could justify for a block that was passing anyway. That constraint is loosening — not because the risk changed, but because the tools that make granular, continuing review economically realistic now exist. "We check the aggregate and trust judgment to catch the rest" isn't gone as an answer. It's weaker than it was five years ago. And that industry-wide effort is a preview of something worth sitting with: the fix that took years of manual, one-by-one work only got built because the rate environment forced it. It's a much better position to build that habit before something forces it than during.

What this actually asks of insurance executives

I'm not offering a view on hedging strategy or reserve methodology — that's not my lane. What is squarely in my lane: whether the organizational conditions exist for the people who already own assumption review to work well with AI — not defer to it, not ignore it, but genuinely partner with it.

• The regulator's own posture is worth studying less for compliance value and more as a design pattern — and it's only become buildable as standing practice, rather than crisis response, recently. Assumption review has always had owners. What it hasn't had is a way to run continuously rather than annually or on trigger. That's what AI now makes affordable — not a replacement for an actuary's judgment, but a way to test a guarantee, a lapse assumption, a legacy block, far more often than a person could manage alone.
• The harder question is whether the talent is being built for that partnership. Treating an AI system's output as the answer is one failure mode. Dismissing what it surfaces because it doesn't match what the model has always said is the other. The judgment in between — knowing when to challenge an AI finding, when to trust it, when to go dig further — isn't automatic. It has to be cultivated, the same way any other analytical judgment is.

Picture your next executive meeting on NAIC's AI Systems Evaluation Tool — whenever the Fall National Meeting takes it up, this becomes an agenda item somewhere. The easy question in that room will be whether you're ready for the exam. The harder, better question is whether the people who own your assumption reviews are equipped to work alongside AI — able to use what it surfaces, and just as able to push back on it when something doesn't add up.

The guaranteed-rate annuity problem wasn't solved by finding more talent — the industry had no shortage of it. It was solved reactively, once a crisis forced the question, and it took years longer than it should have because nobody was set up to ask it before the crisis made it unavoidable. That's the real choice in front of insurance leadership now: build the organizational conditions and the talent to ask that question all the time, with AI doing part of the looking and people doing the judging — or wait for the next rate environment, or the next examiner, to force the question again.

The Call

At 6:42 a.m. on a Tuesday morning in early February, Michael Grant's phone rang.

He was already awake. Public company CEOs rarely sleep well during earnings week.

The caller ID displayed David Ellis, Meridian Commercial Insurance's chief financial officer. David never called before seven unless something had gone wrong.

Michael answered immediately.

"We have a reserve problem."

The independent actuarial review had concluded the previous evening. After weeks of analysis and challenge sessions, the conclusion was unavoidable. Meridian would need to strengthen its commercial casualty reserves by approximately $500 million.

The financial implications were immediate.

Before the markets opened, the board would need to be informed. Rating agencies would request meetings. Investors would expect explanations. The earnings call scheduled later that week would no longer focus on operating performance. It would focus on credibility.

Michael listened quietly before asking a single question.

"How did we not see this coming?"

It was the question every executive, director, analyst, regulator, and shareholder would eventually ask.

It was also the wrong question.

The better question was this:

How did a long sequence of individually reasonable executive decisions become a $500 million reserve development?

The answer did not lie in a single actuarial assumption or one poor management decision.

It began nearly three years earlier.

Three Years Earlier

Three years before the reserve strengthening, Meridian looked exactly like the kind of carrier executives aspire to build.

The regional P&C insurer had earned a reputation for disciplined underwriting, conservative reserving, and consistent financial performance. Reserve development was generally predictable. Analysts viewed the company as steady. Rating agencies considered it well managed. Internally, executives regarded reserve discipline as one of Meridian's competitive strengths.

The quarterly reserve committee reflected that culture.

Led by the chief financial officer, the committee brought together senior leaders from actuarial, claims, underwriting, finance, and risk. Each function contributed a different perspective before management selected the company's carried reserve position.

The chief actuary presented reserve indications based on established actuarial methods and independent reserve reviews.

Claims executives discussed emerging litigation trends, settlement behavior, and developments within the largest claims.

Underwriting reviewed changes in portfolio composition, policy limits, pricing discipline, and business mix.

Finance evaluated the implications for earnings, capital adequacy, and external financial reporting.

The discussions were disciplined.

Assumptions were challenged.

Alternative interpretations were debated.

No one rushed to conclusions.

Looking back, there was remarkably little to criticize.

The organization had experienced executives.

It had sound governance.

It had reliable data.

It had established actuarial methods.

It had exactly the reserve process most well-managed carriers would expect to have.

And yet, within three years, those same executives would be explaining a $500 million reserve strengthening to the board and to the market.

The explanation would not be weak governance.

Nor would it be poor judgment.

It would be something far more subtle.

The organization possessed the information it needed.

It simply possessed it in pieces.

The First Signals

The first indications that something was changing did not arrive dramatically.

They appeared as routine operational observations.

During one quarterly reserve committee meeting, the chief claims officer noted that commercial auto bodily injury claims in several jurisdictions were remaining open longer than historical experience suggested. Settlement negotiations appeared less predictable. Litigation was becoming more complex in certain venues.

The observation attracted discussion but not concern.

Claims professionals encounter fluctuations like these regularly. Individual developments often reflect temporary operational factors rather than lasting changes in the underlying claims environment.

The committee agreed that the trend deserved continued monitoring.

Nothing suggested it justified changing reserves.

The underwriting discussion followed.

The chief underwriting officer described gradual changes in the portfolio. Larger commercial accounts represented a growing share of written premium. Higher liability limits had become increasingly common in several business segments. Competitive market conditions were influencing coverage structures in selected areas.

None of those developments appeared unusual.

Each reflected deliberate business decisions that management had already approved.

The business remained profitable.

Pricing discipline remained intact.

The portfolio continued performing within expectations.

The chief actuary then presented the quarterly reserve indications.

Historical development remained broadly consistent with prior analyses. Multiple actuarial methods continued supporting management's reserve position. Independent actuarial review identified no material reserve deficiency.

Viewed independently, each conclusion was entirely reasonable.

Claims observed subtle operational changes.

Underwriting observed gradual portfolio evolution.

Actuarial observed reserve adequacy supported by historical experience.

No individual observation justified changing the company's reserve position.

The committee approved another quarterly reserve selection.

It was a disciplined decision supported by the evidence available at that time.

What no one yet understood was that the organization was beginning to observe different parts of the same emerging story.

The Institutional Blind Spot

Nothing about Meridian's organizational structure was unusual.

Like most successful P&C insurers, it was organized around specialized functions.

Claims focused on claims.

Underwriting focused on risk selection and portfolio quality.

Actuarial focused on reserve adequacy.

Finance focused on earnings and capital.

Each function was exceptionally good at answering its own questions.

That specialization represented a strength.

It also created an unintended limitation.

Reserve adequacy is not fundamentally a claims question.

Nor an underwriting question.

Nor solely an actuarial question.

It is an institutional judgment that depends upon integrating evidence originating across the enterprise.

During periods of stability, that distinction rarely matters.

Historical experience aligns closely with current operations.

Independent functions naturally reinforce one another.

But when the operating environment begins changing faster than historical experience can reveal, something important happens.

Claims professionals may observe changing litigation behavior months before those developments become statistically visible.

Underwriters may recognize shifts in portfolio characteristics before those changes influence reserve development.

Finance may begin asking different questions about capital while actuarial analyses remain appropriately anchored in historical data.

Each observation is valid.

Each reflects disciplined professional judgment.

Each remains incomplete.

The reserve committee brought those perspectives together.

It integrated professional opinions.

It did not necessarily integrate the underlying evidence.

That distinction explains how well-governed institutions can still reach consequential decisions with an incomplete understanding of emerging enterprise risk.

Meridian was not suffering from a lack of information.

The organization knew many important things.

It simply did not yet know what those things meant when viewed together.

Rewinding the Story

Now rewind the story.

The portfolio is the same.

The claims are the same.

The executives are the same.

The reserve committee is the same.

The governance process is unchanged.

No authority has been delegated to AI.

No actuarial methods have been replaced.

No reserve decisions have been automated.

Only one thing changes.

The institution begins making reserve decisions with a more complete understanding of the evidence already available across the enterprise.

Instead of reviewing information that has been organized independently by claims, underwriting, actuarial, finance, and risk, the reserve committee begins with an integrated view of the emerging risk environment.

That distinction may appear subtle.

It is anything but.

In the original timeline, each function accurately described what it was observing.

Claims reported longer settlement cycles.

Underwriting described a gradually changing portfolio.

Actuarial presented reserve indications that remained within established ranges.

Finance evaluated capital and earnings implications.

Every presentation was accurate.

Every conclusion was professionally sound.

What the committee never saw was how those observations were beginning to reinforce one another.

Now imagine the same meeting supported by an AI-enabled decision intelligence capability.

Before the committee convenes, the system continuously evaluates information flowing across the enterprise—not to make reserve recommendations, but to identify relationships that deserve executive attention.

It examines claim notes, case reserve movements, settlement duration, litigation activity, jurisdictional trends, changes in policy limits, shifts in portfolio composition, external legal developments, and traditional actuarial analyses.

Its purpose is not to predict ultimate losses.

Its purpose is to answer a different question:

Are independently observed developments beginning to describe the same emerging enterprise risk?

Rather than producing another dashboard, the system highlights combinations of evidence that would otherwise remain separated by organizational boundaries.

Claims managers have independently observed increasing settlement duration within several commercial auto segments.

Underwriting has independently documented continued growth in higher-limit policies within many of those same segments.

External litigation data suggests increasing plaintiff success in overlapping jurisdictions.

Traditional actuarial indications remain within acceptable ranges, but recent development is gradually migrating toward the upper end of those ranges.

None of these observations is individually conclusive.

Together, however, they justify asking different questions before the next reserve decision.

Notice what has not changed.

The chief actuary still owns the actuarial analysis.

Claims leadership still evaluates claim behavior.

Underwriting still assesses the portfolio.

Finance still considers capital implications.

The reserve committee still debates.

Management still decides.

The board still exercises oversight.

AI has not replaced professional judgment.

It has strengthened the evidence supporting professional judgment.

Instead of asking,

"Do today's reserve indications remain reasonable?"

the committee begins asking,

"Does our historical development fully reflect what the rest of the organization is already beginning to observe?"

That is a fundamentally different conversation.

The committee is no longer evaluating isolated evidence. It is evaluating the emerging enterprise narrative that evidence collectively describes.

The Decision Changes Before the Outcome Does

The first visible change at Meridian is not a reserve adjustment.

It is the discussion preceding the reserve decision.

During the next quarterly review, actuarial indications continue supporting management's reserve position.

Historically, that would likely have concluded the discussion.

Instead, it becomes the starting point.

Claims leadership is asked whether the changing settlement patterns represent temporary operational variation or an emerging shift in litigation behavior.

Underwriting reviews whether the evolving portfolio characteristics are concentrated within specific industries, geographies, or coverage structures.

Actuarial evaluates additional scenarios while continuing to rely upon established reserving methods.

No one proposes a $500 million reserve increase.

Nor should they.

The available evidence still supports management's selected reserve position.

What changes is the institution's understanding of uncertainty.

Over subsequent quarters, as additional claim development enters the actuarial analyses, the findings no longer appear unexpected.

Management has already been investigating the underlying drivers.

Reserve strengthening still occurs.

Long-tail casualty business remains uncertain.

No analytical capability eliminates that uncertainty.

The difference is that reserve adjustments occur progressively as the organization's understanding evolves rather than emerging as a single material financial event.

Capital planning becomes more measured.

Investor communication becomes more predictable.

Management discussions become more forward-looking.

The financial impact remains. The surprise does not.

Beyond Reserving

Reserve development illustrates a broader principle that applies throughout P&C insurance.

Every consequential decision is made before uncertainty has been eliminated.

Underwriting decisions commit capital before future losses are known.

Claims decisions resolve complex situations before every fact has emerged.

Reinsurance decisions reshape retained risk before the next catastrophe occurs.

Reserve decisions recognize liabilities before ultimate claim outcomes become fully visible.

The common challenge is not technology.

It is decision making under uncertainty.

That is why the executive conversation around AI should begin to change.

Today's discussion often centers on models, copilots, agents, orchestration platforms, and automation.

Those capabilities matter.

They improve productivity.

They streamline workflows.

They reduce operating expense.

But those are implementation capabilities.

Enterprise value is created somewhere else.

It is created when institutions consistently make better consequential decisions.

That distinction changes how executives evaluate AI investments.

Instead of asking,

"How many AI capabilities have we deployed?"

they begin asking,

"Which consequential decisions have become measurably better because of them?"

That is ultimately the only business question that matters.

The Decision Before the Decision

Meridian's story is not fundamentally about reserving.

It is about institutional decision making.

The company did not experience a $500 million reserve strengthening because it lacked capable executives, disciplined actuarial methods, or sound governance.

It experienced one because individually reasonable observations remained institutionally disconnected until history had already confirmed what operations had begun to signal.

That distinction reaches well beyond reserving.

Every consequential underwriting decision reflects the institution's understanding of risk at a particular moment.

Every significant claims decision reflects its understanding of liability, customer impact, and financial exposure.

Every reinsurance decision reflects management's understanding of the portfolio before uncertainty has been resolved.

In each case, executives must commit capital before perfect information exists.

The quality of those decisions depends upon the quality of the evidence available before the commitment is made.

That is where AI has the opportunity to reshape P&C insurance.

Not by replacing executive judgment.

Not by automating accountability.

Not by predicting the future with perfect accuracy.

Its greatest contribution is enabling institutions to assemble, interpret, and challenge evidence earlier, allowing experienced executives to make better-informed consequential decisions while they still have the opportunity to influence the outcome.

For decades, reserve committees have begun with a familiar question:

"Based on the evidence before us today, what reserve should we establish?"

The AI era introduces an earlier and arguably more important question:

"Have we assembled every meaningful piece of evidence the institution already possesses before making one of its most consequential financial decisions?"

The institutions that outperform in the AI era will not necessarily be those that deploy the most AI.

They will be the ones that consistently assemble better evidence before making consequential decisions.

In the AI era, competitive advantage will belong less to the institutions that possess the most information than to those that assemble the best evidence before making consequential decisions.

As soon as I heard about Flock, the now-huge network of cameras that can automatically read license plates and instantly report their location to authorities, I wondered how long it would take before we read about police officers using the network to track girlfriends, wives and exes. 

It actually took longer than I thought, but here we are: The Washington Post reported a week ago that at least 50 officers throughout the U.S. have been accused of stalking by using Flock and similar systems. All sorts of national media have now followed the Post story, and we're in an environment of heightened sensibilities about surveillance and about how scary AI might turn out to be, so this story isn't going away any time soon. 

In the swamp that is today's social media, some conspiracy theorists are trying to drag in the insurance industry, saying it uses Flock cameras in nefarious ways. As the scandal unfolds, we need to not only make sure we avoid any unethical uses of camera networks but can respond quickly and convincingly to any online trolls alleging abuse.

Let's have a look. 

Many years ago, a good friend was driving an old car in Washington, DC, when he stopped at a stop sign and saw a pedestrian motioning to him that he needed directions. Tim rolled his window down, and the guy stuck a gun in his face, ordering him out of the car. But the car's clutch was dying, and the carjacker couldn't figure it out. He lurched a few feet and stalled. Another few feet and another stall. He stalled a third time before he even made it halfway through the intersection. He got out of the car, threw the keys into Tim's chest and ran off.

If the carjacker had succeeded and been in modern DC, he would have been a perfect candidate for Flock. Tim could have simply told a 911 dispatcher what his license plate was, and Flock's 120,000 cameras throughout the U.S. (except Alaska) would have instantly been on the lookout. Flock works with 7,000 police departments, or 40% of the total in the U.S., so once its AI spotted the license plate on the stolen car it would have been able to alert police and have a pretty good chance of recovering it. 

In theory, Flock will gradually prevent more and more thefts, because thieves will either be caught and incarcerated or will simply realize that car theft is a losing proposition.

Somewhere in your agency right now, someone is probably matching carrier statements by hand, and by the time they finish, the numbers they're checking are already out of date.

That's direct bill reconciliation, one of the most manual, error-prone jobs in an agency's back office. Someone matches carrier statements against your management system line by line, hunting for the discrepancies that always show up: a payment posted to the wrong policy, a commission rate that doesn't match the schedule, a premium finance installment that's out of sync, a chargeback nobody flagged. It's tedious work, and for years it's simply been accepted as the cost of getting paid what you're owed. The book grows, the statement volume grows with it, and reconciliation stays exactly as manual as when the agency was half the size.

Here's the argument I'd make to any operations leader reading this: reconciliation isn't a back office chore anymore. It's one of the highest leverage places in your entire financial operation to put AI to work, and the shift isn't incremental. It's closer to what personal computing did to batch processing: things that used to be periodic events, the month end close, the quarterly true up, are starting to happen continuously instead. And when that happens, the knowledge of where the agency actually stands stops living in one person's spreadsheet and starts belonging to the agency itself, in real time.

What manual review actually misses

Ask anyone who's done reconciliation by hand what keeps them up at night, and it's rarely the errors they catch. It's the ones they don't. A commission underpayment on a single policy is a rounding error. The same pattern running quietly across a book of business for two quarters is a real number, and by the time someone notices, it's already happened.

AI-powered reconciliation catches a different category of error than manual review, not because it's more careful, but because it's built to look for patterns rather than line items. It flags a commission rate that's drifted from the schedule across dozens of policies at once, not just the one a reviewer happened to pull. It catches a payment landed against the wrong policy before it compounds into a billing dispute. And increasingly, it catches the kind of discrepancy that turns into an E&O claim: a missed cancellation, earned premium never reconciled, a client charged for coverage that had already lapsed. Manual review finds what someone thinks to look for. Automated reconciliation finds what's actually there.

Why this is the highest leverage place to automate

Agencies have no shortage of processes worth improving, and it's tempting to assume the highest value automation lives somewhere flashier, in underwriting or the sales motion. Reconciliation rarely makes that list, and that's exactly why it's underrated.

1. It touches every transaction, not just the exceptions. Automating a workflow that only runs when something goes wrong helps at the margins. Reconciliation runs on every payment, every commission, every policy in the book, so improving it once compounds across the entire book every month.

2. It's where money and data are supposed to agree, and usually don't. Every other financial operations problem, delayed close, unclear profitability, cash flow surprises, traces back to a reconciliation gap somewhere upstream. Fix the reconciliation and a lot of the downstream noise quiets down on its own. It also means the agency, not one person's manual process, becomes the one holding the accurate picture.

3. It's measurable in a way a lot of operational improvements aren't. You can point to the hours recovered, the discrepancies caught before they became client issues, the days shaved off the close. The close either happens faster and cleaner or it doesn't, and everyone in the agency can see which, making it one of the easier automation investments to justify.

What the recovered time actually looks like

The honest answer is that most agencies underestimate how much staff time direct bill reconciliation consumes until they see it recovered. It's not unusual for a single person to spend a meaningful share of every month matching statements by hand, time an agency pays for that produces no new business and serves no client directly. It just keeps the lights on.

The operations leaders getting this right aren't redeploying that time into more reconciliation, done faster. They're redeploying it into the work reconciliation was always supposed to make room for: catching a renewal about to lapse, following up on an account that looks profitable on paper but isn't, having the conversation with a producer about which accounts are actually worth the effort. The close stops being a monthly fire drill and starts being a byproduct of work already happening in real time.

The month end close was never the point

The month end close exists because reconciling by hand takes time, and batching the work into a monthly cycle was the only way to make it manageable. Take away the reason for the batch, and the batch stops making sense. That's the real shift underway: not that reconciliation gets faster, but that a monthly close as a distinct event starts to look like an artifact of how the work used to get done.

Agencies that treat this as a technology upgrade will get a faster close. Agencies that treat it as a chance to see their own numbers continuously, rather than waiting for someone to hand them a reconciled report two weeks after the fact, will get something more valuable: the agency itself, not whoever happens to be doing the matching that week, becomes the one who actually knows where things stand. That's worth more than the hours it saves.

The insurance industry is facing a hard question: Is the way we work actually working? 

For many, the answer is no.

Disconnected workflows remain one of the biggest pain points across insurance operations. Quotes bounce between departments. Claims pile up in inboxes. Approvals disappear into email threads. Every manual handoff creates friction, slows resolution times, and chips away at customer and broker relationships. 

Guidewire, Duck Creek, and other core systems have made real strides, but the deeper issue is that integrations fall apart because the underlying workflows are outdated, inefficient, poorly documented, and full of exceptions that no one has mapped out.

The fix is not simply adding new people or platforms; it's streamlining and standardizing workflows first and then using that enhanced foundation as the platform for making informed workforce and technology decisions.

The Standardization Gap No One Talks About

Before any technology conversation can happen, carriers have to confront the problem at a more fundamental level by addressing inconsistent processes across their own organizations.

Most insurance organizations have established ways of doing things, and many of them work passably well. But having a "working" process and one that's optimized are two different matters. Seemingly, there's always room to tighten handoffs, eliminate process redundancies, and create consistency across teams without overhauling what already works. Nevertheless, even rock-solid workflows benefit from a fresh look–because what worked five years ago is not ready for today's work volume and complexity, or available technology options.

The Capgemini World Property and Casualty Insurance Report 2026 confirmed that even carriers that are heavily invested in AI are dependent on workflows built for human execution. For these businesses, the remaining challenge slowing process redesign for them lies in solving gaps in collaboration and data readiness. In plain terms: technology cannot fix a process. It can only accelerate it, for better or worse.

Standardization means thoroughly understanding and documenting how work actually moves through the organization, from submission intake to policy issuance to claims resolution, and then deciding how it should move. This is tedious, time-consuming work, but doing it pays huge dividends in the long run – and carriers that skip it and jump straight to automation consistently struggle to connect their technology investments to measurable returns.

Complexity Runs Far Deeper Than Most Carriers Admit

Handling non-standard or specialty risks exponentially increases the problem and raises the stakes.

For example, in 2024, excess and surplus lines markets surpassed $81 billion in written policies, a 12% jump over the previous year, and the continuation of a five-year trend of double-digit growth. This expansion brings volume, and with it, further exposure of manual processing's inherent vulnerabilities.

The compliance burden alone illustrates the problem clearly. According to the Wholesale and Specialty Insurance Association's 2025 Compliance Benchmarking Report, the most common data filing error – incorrectly applying tax rates – accounts for 38% of all rejections, followed by missed deadlines at 29%.

These numbers aren't the result of technology failures but instead stem from undocumented processes performed manually and varying from person to person – call it "vibe underwriting," if you will. Fixing the first problem, incorrect application of prevailing tax rates, does not require purchasing a workflow automation platform. It requires deliberately mapping and documenting every step of the process and eliminating or correcting the inconsistencies that have built up over time.

Undocumented workarounds and informal "tribal knowledge" are among the most common causes of errors in compliance-heavy workflows, and no software or LLM can solve this until the process itself is defined and accountability for consistency is assigned.

Standardization in specialty lines does not mean stripping out flexibility. It means building structured, auditable workflows wherever possible, so underwriters and brokers spend their time on judgment calls that actually require expertise, not on chasing paperwork or correcting errors that could have easily been avoided in the first place.

Where Technology Fits In

Artificial intelligence and automation are already delivering real results in production. According to McKinsey, leading P&C insurers are seeing 40% reductions in underwriting costs through automation and predictive analytics. In specialty lines, around 52% of insurers have deployed AI tools for faster underwriting, and 47% have improved processing accuracy and turnaround times using AI in their claims workflows.

In claims specifically, straight-through processing is becoming the standard outcome for routine losses, reducing cycle time and manual touchpoints across the lifecycle. With AI-automated data extraction, document review, compliance checks, and routing, adjusters are able to focus on complex, potentially costly losses needing human evaluation.

Low-code automation platforms are emerging as the practical path forward for managing high-variability submissions. These tools let business users build and update workflows without waiting on development resources, even as appetite, rating logic, and product lines constantly shift to meet dynamic markets. Cloud-native, API-connected platforms add another layer of flexibility, providing scalability to grow capacity during CAT events and other demand surges, and scale back during quieter periods without adjusting permanent headcount.

Standardized Processes Produce Better Staffing Decisions

This is the critical juncture for workflow efficiency and workforce strategy. You cannot right-size a team around a broken or inconsistent process.

If a claims department runs on manual data entry, inconsistent documentation standards, and siloed systems, staffing levels will reflect that dysfunction. Bring more hires into a fragmented workflow and you get a bigger, but still fragmented workflow. Standardize the process first, then determine how many people you need and what skills those roles actually require.

Successful automation implementations in insurance have shown up to a 50% reduction in employee workload alongside a 25% to 35% increase in team productivity. To yield maximum value from these gains requires an additional step of redeploying human experts, rather than simply reducing headcount. After all, AI is a tool, and you wouldn't replace a skilled plumber with a wrench.

When repetitive, high-volume tasks are automated, underwriters underwrite. Adjusters focus on complex losses. And every role across the insurance value chain is empowered to focus on building broker and agent relationships instead of spending time on tasks that can be automated with human oversight.

A growing number of carriers are also rethinking their workforce structure entirely, blending lean internal leadership teams with technology and specialized vendor partners. This model lets organizations scale up during peak periods without carrying excess fixed overhead year-round.

A Practical Starting Point

The path forward does not require transforming every process at once.

Start with a workflow audit to isolate the areas generating the most operational friction. That typically means FNOL intake, renewal workflows, compliance reporting, and submission intake.

Map what actually happens today, not the idealized version your business may have documented sometime in the past. And do not just talk to the people doing the work. Engage them fully in the mapping process itself. They are the subject matter experts, and this is an opportunity to recognize and create a competitive advantage from their contributions. The people closest to the work know where the friction is, where the workarounds live, and precisely where the process breaks down under pressure. Bringing them into the solution builds ownership, surfaces insights no audit tool will catch, and signals that you value your team for their expertise, and not just their output.

From there, define the standard. Map what the process should look like, step by step, with ownership at each stage clearly defined. Only then are you prepared to evaluate technology platforms, choose the tools that best integrate with existing systems and give managers real-time visibility into where work stands.

Perhaps most important: train your people on the new workflows in their entirety, and not the new software. Track results with concrete metrics (from set baselines): processing times, error rates, submission-to-bind cycle times, and compliance rejection rates. Adjust the new workflows as these data direct.

The Bottom Line

Process efficiency is not an IT initiative. It is a discipline combining business workforce strategy. Insurers that audit their workflows, standardize their operations, and then layer in the right technology will hire smarter, retain better, and serve clients and brokers faster.

The carriers gaining ground right now are not just doing more. They are doing more with less friction. That starts with knowing exactly where friction lives, and having the discipline to fix it before committing to your next platform.

Sources: Capgemini World Property and Casualty Insurance Report 2026; McKinsey and Company; Wholesale and Specialty Insurance Association 2025 Compliance Benchmarking Report; Datagrid Workflow Automation Analysis; Patra 2025 AI and Insurtech Trends Report