Download

Friday the 13th Comes for Our Tech

The air traffic debacle and email outage on Friday don't just call for more careful management of our technology. They also offer a stark warning. 

Image
Insurance

When I ran the editing operation for the Wall Street Journal's European edition in the '80s, the "lockup" at midnight (or so) was frequently so chaotic that I posited what I called Carroll's law. Building on Murphy's law – everything that can go wrong will go wrong – Carroll's law said Murphy was an optimist.

I'm feeling déjà vu, given all the technology-related chaos of the past week, centered on Friday the 13th, when all U.S. air traffic was shut down for hours and Microsoft's Outlook was only intermittently operating after a full-on outage the day before, for what amounts to the email backbone for the world economy.

As much as there are individual reasons for those outages and other recent ones, the problems point to a pattern. The transition we're going through to a completely wired, AI world is so fundamental that two things are true: 

1) We need to be a lot more careful about work like the upgrades to the FAA system, and we can head off a lot of problems if we are.

2) There will still be chaos from time to time, including for the insurance industry, especially as we start to roll out AI agents. So buckle up.

Let's have a look.

I first started thinking a couple of weeks ago about writing about the risks that arise from connecting everything to everything through the Internet of Things. The impetus was a wave of attacks on refrigeration systems throughout Europe – including in France, the U.K., Italy, Belgium, Sweden, Luxembourg and Estonia – followed by similar attacks on commissaries at U.S. military installations. The hackers raised the temperature inside 900 freezer units across 17 countries to as high as 120 degrees Fahrenheit, causing millions of dollars of food to go bad overnight and be thrown out. 

France, which suffered two-thirds of the attacks on freezers, blamed Iran, and geopolitical tensions could well lead to such attacks. North Korea, Russia and China all have active military cyber units. 

It's also possible that teenagers who in the past might have spraypainted graffiti on a school wall will now use AI to indulge their anger and show off in much more destructive ways. 

In any case, the cyber attacks show that the power of the IoT comes with a potential downside. Connecting essentially every device to every other device gives hackers more routes into systems they want to attack. And industrial systems, such as those that let supermarkets monitor and manage refrigeration centrally typically aren't built to the same security specifications that the main IT systems are.

Insurers will have to be alert both about clients and about themselves. Clients may suffer theft, business interruption, claims of negligence against directors and officers, and so on. Insurers may face attacks themselves as they link to more sensors, third-party data sources, and providers of digital services such as payments. Insurers certainly make juicy targets, given all the important information they have on individuals and businesses. 

The Wall Street Journal published a story with the headline, "Supermarket-Fridge Hacks Show Risk of Linking Everything to the Internet", that said: "Artificial intelligence suddenly makes the problem much bigger, because hackers are automating their hunt for exposed web connections and running searches at blistering speeds." The article adds that 20% of IoT devices already have at least one known cybersecurity vulnerability. 

It turns out, though, that the refrigerator cyberattacks were just the start. On Thursday, bleeding into Friday, Microsoft's Outlook email system went down. As Wired wrote in a story headlined "The Microsoft Outage Is a Warning for the World": "The outages have caused widespread disruption to schools, courts, governments, and law enforcement agencies—plus countless businesses and millions of employees—around the globe, highlighting how many of the world’s most crucial systems rely upon technologies from a small number of tech companies. And when they fail, the ramifications can be huge."

Full details on the cause of the outage have yet to be provided. At the least, it shows the danger of concentration for a key service such as communication. We've actually seen even more, though smaller, examples of dangerous concentration just since Friday. United Airlines, which controls a major chunk of air travel in the U.S. and internationally, had to delay all departing flights for more than an hour and a half on Monday because of a software issue.  A technical glitch with software at Chase Bank delayed online banking and caused transaction glitches for both itself and other major banks.

Concentration will only increase from here, so we need to be prepared for the fact that a single problem can cause widespread trouble.

There is some speculation that AI may have played a role in the Outlook trouble, given that Microsoft laid off 4,800 employees in July and just offered voluntary buyouts to 60,000 people, or 27% of its workforce, while saying the company is turning more coding work over to AI. 

Which brings us to the FAA communications disaster that shut down air travel for five hours in the U.S. on Friday. The current official explanation is that a single, old router in Arkansas was overwhelmed during an upgrade to new technology, but some officials initially said an upgrade to AI technology known as SMART may also have been involved. Oddly, the disagreement is very public and at the cabinet level  – the secretary of Transportation insists the AI work had no bearing on the outage, while the secretary of Homeland Security has suggested at least twice that the AI initiative may have been to blame.

Whatever the exact problem, the FAA issue should be the simplest to avoid. It just requires some basic hygiene. For instance, you have to give yourself a way to instantly undo an upgrade if it starts to go south. And whatever you do, you don't do a major upgrade during a busy period -- such as a Friday afternoon, when millions of people are flying somewhere for the weekend. You do the upgrade in the middle of the night, when a problem can be corrected before much of anyone notices. 

But people will ignore basic principles of software management, just as they'll be too reliant on single sources of key digital services such as communication and will be too slow to tackle the vulnerabilities that come from connecting to systems and services that aren't secure enough. And AI will greatly accelerate the pace at which problems surface. So we should consider recent events as warnings of future problems, not as one-offs. 

There is good news in the short term, though, for those of you who are superstitious types. Friday wasn't just the 13th; an astrology-curious acquaintance tells me Mercury and Venus both ended their retrogrades that day, bringing some supposed relief after weeks of celestial troublemaking. (Mercury, I'm told, can be especially problematic. It's blamed for everything from miscommunications to travel mishaps.) But both planets have resumed their normal forward movement across the sky, so astrologers give us an all clear – or at least a mostly clear, because the planets' "shadow" periods will last for a few weeks.  

Cheers,'

Paul

P.S. I later learned, courtesy of Google, that many people had beaten me to the Carroll adaptation of Murphy's law. Oh, well. 

P.P.S. My favorite tweet on the tech outages: 

"If you booked your United Airlines flight through your Chase Bank account and need to access those records on your Outlook, what do you do?"

The Flaw in 'Human-in-the-Loop'

An HBR article argues that human supervisors quickly become unreliable as the AI becomes increasingly accurate.

Image
Insurance

The notion that there must always be a "human in the loop" has become de rigueur in any discussion of AI implementation in business. The cynical side of me suspects that part of the reason is to reassure everyone that AI won't lead to mass layoffs, but there are also plenty of practical reasons to have humans check the work of AIs.

Or maybe not. At least not for long.

A recent article in Harvard Business Review argues that human supervisors quickly become numb and simply approve the recommendations of the AI as it becomes increasingly accurate. And I think the author has a point, based on what I've seen of human/AI interactions in semi-autonomous cars. 

There needs to be a better way.

Let's have a look.

Early in the development of its driverless cars, Google gave demo models to employees. Google warned them that the AI wasn't perfect and instructed them to have their hands on the wheel, always ready to take control. That worked... for a little while. A very little while. 

As the AI proved to be reliable, the drivers began to relax their hold on the wheel, then started to take their eyes off the road, then began reading or looking at their phones. When the car signaled that the drivers had to take over, they needed perhaps 10 seconds to fully reengage. Given that a car going 60mph travels nearly 300 yards in 10 seconds, Google realized its hybrid approach wouldn't work and committed to producing only cars that could be fully autonomous. 

Other makers of autonomous vehicles, most notably Tesla, continue to offer cars that take a hybrid, human/AI approach, but I think that's irresponsible (as I wrote at length a year ago). 

Now, an article in Harvard Business Review offers a similar view about the dangers of having a "human in the loop" to supervise work done by AIs. The author, a senior fellow at Cal, writes:

"Decades of research on automation show that better automation systems lead to worse human safety layers; as automation grows more reliable, the people monitoring it grow complacent, their skills atrophy, and they become less attentive.... AI compounds the problem. Its failures sound as confident, fluent, and well-structured as its successes....

"Even when validation works and people catch errors, every correction is captured as training data. Each exception a person handles teaches the system to handle that exception; the better the data, the smaller the scope of human validation needed in the future."

The article says the lack of reliability of the humans in the loop is actually just the start of the problem. He writes:

"The larger problem with the human-in-the-loop organization, though, is that the entire blueprint reads the transformation backwards. It organizes the company around the shrinking frontier of what machines can’t yet do. But there’s a larger opportunity here. Because AI removes constraints, it enables new combinations of work, new decision rights, and new forms of value. Organizations, accordingly, have an opportunity to reorganize and 'rebundle' around those new decisions rather than the shrinking residue of the old work."

He goes on to lay out how to think about rebundling: "Instead of asking which tasks the machine can take over, leading organizations should ask which new roles the technology makes possible, and then redesign the organization around them." He lists several steps that can lead you to new approaches, which seem intriguing and which you may want to explore.

For now, though, I just want to make the point that "human in the loop" isn't a panacea, certainly not in the long term. The reviewers of AI work can quickly become as reliant on the AI as those Google employees who found themselves with their hands off the wheel, their eyes off the road and their minds far away from driving when a problem popped up in front of them.

Cheers,

Paul

Independent Insurance Agencies in 2030

Independent insurance agencies face mounting operational inefficiencies, but agentic AI could unlock $4.7 billion in productivity gains by 2030.

Insurance

Independent insurance agencies, the backbone of personalized coverage in the U.S., are under siege from a perfect storm of operational inefficiencies, economic headwinds, regulatory shifts, environmental risks, and relentless consolidation. From 2022 to 2025, agencies grappled with inefficiencies costing 20-30% of annual revenue—translating to $200,000-$300,000 in lost profits for a mid-sized firm generating $1 million in premiums. Key performance indicators (KPIs) like customer acquisition cost (CAC) ballooned to $900 on average, while revenue per agent stagnated at $400,000-$500,000, eroding return on investment (ROI) and return on ad spending (ROAS) as retention rates slid below 85%.

Looking ahead to 2030, these pain points will intensify: climate-driven premium hikes could add 22% to costs in high-risk regions like Florida and Louisiana, consolidation may swallow 20% more independents, and economic volatility from inflation and geopolitics could push expense ratios above 40%. Our forecasts project CAC rising 15-20% annually in coastal states, with revenue per employee dropping 10% due to talent shortages and tech lags.

Yet, hope lies in agentic and generative AI voice and productivity agents like our Evolve AI Agent. Today, these tools slash claims processing time by 40%, boosting productivity and cutting CAC by 25%. By 2030, advances in multi-agent systems and real-time voice analytics will automate 60% of workflows, potentially unlocking $4.7 billion in AI-driven premiums industry-wide and restoring ROAS to 4:1 ratios. 

Pain Points From 2022-2025

The past three years have been a crucible for independent insurance agencies, marked by post-pandemic recovery, inflationary pressures, and accelerating digital disruption. Our proprietary research, corroborated by industry benchmarks, reveals operational inefficiencies as the primary culprit, draining resources and stifling growth. Agencies reported average monthly revenue losses of $16,667 (20% of $1M annual revenue benchmark), equating to $200,000 annually per firm, largely from manual processes and outdated systems.

Operational Inefficiencies and Associated Costs

At the core are legacy workflows: manual policy renewals, claims tracking, and commission reconciliation consume 30-40 hours per agent weekly, per our surveys. A 2023 IDC study pegs these inefficiencies at 20-30% of revenue lost yearly, or $2,000-$3,000 monthly per employee in wasted time (valued at $50/hour). Inefficiencies in distribution—such as siloed data and lengthy sales cycles—exacerbate this, leading to 20% lost revenue overall. For a 10-agent agency, this translates to $400,000 in annual opportunity costs, including negative ROAS from underperforming digital leads (often below 2:1 due to poor targeting).

Customer expectations shifted dramatically post-2022, with 80% demanding digital-first service (PwC). Yet, 67% of executives cited data quality issues as barriers, per Deloitte, resulting in inconsistent reporting and 13-point drops in satisfaction scores. This churn costs agencies $5,000-$10,000 per lost client annually in foregone renewals. Talent woes compound the issue: aging workforces (average age 46+) and Gen Z attrition led to 15-20% turnover rates, per our data, with recruitment CAC hitting $4,000 per hire amid competition from insurtechs.

Catastrophic claims surged 35% from weather events, overwhelming manual processes and spiking loss ratios to 65-70%. Agencies in high-exposure areas like the Gulf Coast faced 80% higher nonrenewal rates, per Treasury data, eroding revenue per policy by 10-15%.

KPI Breakdown: Quantifying the Losses

Our analysis of 2023-2025 benchmarks highlights deteriorating metrics:

KPI

2023-2025

Benchmark

Monthly Loss ImpactInefficiency Indicator
CAC$900 average (up 30% from 2022)$7,500/month for 50 new clients (inefficient channels)High due to manual lead nurturing; 5-7x retention cost.

ROI

(Marketing)

2-3:1 (down from 4:1 pre-2022)$10,000/month in wasted ad spendingNegative from poor data silos; 3-5% lower profitability.
ROAS1.5-2:1$15,000/month lost revenueFragmented tracking leads to 20% over-spending on low-ROI tactics.
Revenue per Agent$400K-$500K$8,333/month per agent shortfallStagnant from 40-hour manual tasks; 10% below industry avg.

Revenue/Cost

per Employee

3:1 ratio (expense ratio

35-40%)

$20,000/month agency-wideHigh payroll (40% of revenue) signals inefficiency.
Retention Rate82-85%$25,000/month in churn (10 clients)Data inconsistencies cause 13-point satisfaction dips.

*These figures, drawn from sources like Insurance Journal and our surveys, underscore a $100B+ industry-wide revenue bleed from 2022-2025, with independents hit hardest due to limited scale.

Challenges Over the Next 60 Months

Projections to 2030 paint a turbulent landscape, with outside factors amplifying pain points. Economic slowdowns (2-3% GDP growth), political deregulation, environmental catastrophes, and M&A frenzy could erode independent market share from 62% to 55%, per Big I data. Our models forecast 15% KPI deterioration in high-risk regions, with monthly losses climbing to $25,000+ per agency.

Economic, Political, Environmental, and Consolidation Factors

Economic Pressures: Inflation stabilizing at 2.7% masks tariff hikes and geopolitical tensions, potentially reigniting claims costs by 10-15% (BCG). Non-life ROE may dip to 11% by 2025 before rebounding, but independents face 3-5% lower growth from high interest rates squeezing small-business clients. Political rollbacks under new administrations could cut compliance costs by 10% but spike volatility, raising reinsurance premiums 20% in trade-dependent regions like the Midwest.

Environmental Risks: Climate events drove $357B in 2023 losses, with a $234B protection gap. Premiums in top-risk counties rose 22% (2020-2023), averaging $500 household jumps. Forecasts to 2030 predict 35% more severe weather, pushing loss ratios to 75% and nonrenewals 80% higher in exposed areas.

Regional Differences: Coastal Southeast (Fla., La., Texas) faces acute adversity: Fla. premiums surged 42% in 2023 to $11K/year, with 13 insurers exiting; uninsurability risks 20% of policies by 2028, per Treasury. Gulf states could see CAC +25% from fraud spikes post-disasters. Midwest hail-prone areas (e.g., Minn.) face 15-20% premium hikes, eroding revenue per agent by 12%. West (Calif., Colo.) grapples with wildfires, with 82% higher costs in high-risk ZIPs ($2,321 avg. premium). Rural Heartland agencies, less climate-hit, suffer more from consolidation (250% M&A rise since 2013), losing 15% market share to PE-backed giants.

Consolidation Wave: M&A hit record highs in 2024 (third-highest volume), with PE driving 17 new $1B+ brokers. By 2030, 20% more mid-sized independents ($500K-$10M revenue) may consolidate, per MarshBerry, slashing contingent commissions 10-15% and pressuring ROAS via lost carrier alignments. Political stability aids PE "dry powder," but independents in urban Northeast (N.Y., Mass.) face 30% higher acquisition pressure vs. rural South.

Projected KPI Deteriorations
KPI2025-2030 ForecastMonthly/Annual Loss ProjectionRegional Variance
CAC+15-20%/year$11,250/month (coastal); $9,000 (inland)+25% Southeast from claims surge.
ROI1.5-2:1$15,000/month wasted -10% Midwest from tariffs. 
ROAS1-1.5:1$20,000/month-15% West from wildfire nonrenewals.
Revenue per Agent$350K-$450K (-10%)$10,000/month/agent-12% Gulf from talent flight.
Revenue/Cost per Employee2.5:1 (expense 40-45%)$25,000/month

+18% rural from M&A

competition.

*These trends, informed by Deloitte and EY outlooks, signal $150B in cumulative losses unless addressed.

Agentic and GenAI Voice and Productivity Agents

Agentic AI—autonomous systems orchestrating workflows—and generative AI (e.g., voice-enabled chatbots) offer a lifeline, automating 40-50% of tasks today and scaling to 60% by 2030.

Current Benefits for Agency Owners

Voice AI agents handle inbound calls, triaging claims in minutes (40% faster), reducing CAC 25% via automated lead qualification. In our trials, agencies saw 20% cross-sell uplift from personalized scripts, boosting revenue per client $500/year. Fraud detection flags anomalies in real-time, cutting losses 15-20% ($5,000/month savings). Productivity soars: agents reclaim 20-30 hours/week for high-value sales, lifting revenue per agent 15%. ROI hits 5:1 on implementation, with ROAS improving to 3:1 through data-driven targeting.

For operations, multi-agent systems integrate CRM and policy tools, slashing expense ratios 10% by minimizing silos. Customer retention climbs 10-13 points via empathetic voice interactions, per Salesforce.

Future Benefits and Technological Advances (2025-2030)

By 2030, AI premiums could reach $4.7B annually (Deloitte), with voice AI evolving via multi-agent orchestration and Retrieval-Augmented Generation for hyper-personalization. Advances include real-time risk profiling (seconds vs. weeks), dynamic pricing via IoT/voice data, and predictive claims prevention—reducing loss ratios 20%. Agencies will automate 50%+ underwriting, freeing staff for strategy and cutting CAC 30-40%.

In high-risk regions, voice agents will enable resilient quoting amid disasters, stabilizing revenue per employee +15%. Overall, expect 10-20% premium growth, 20-40% onboarding cost cuts, and 65% complaint reductions (McKinsey).

Conclusion and Call to Action

Independent agencies stand at a crossroads: cling to outdated models and face $25K+ monthly erosions, or embrace AI to reclaim efficiency and growth.

Sources and Citations:

1. Renaissance Alliance: Challenges in 2025

2. Igloo Insure: Hidden Costs

3. Indio: Operational Inefficiencies

4. Centric Consulting: Industry Challenges

5. ASNOA: KPIs for Owners

6. InsightSoftware: Insurance KPIs

7. Deloitte: 2025 Outlook

8. Agent for the Future: 2025 Predictions

9. BCG: Insurance 2025

10. AgileCap: Political Impact

11. PMC: Climate Risks

12. Treasury: Homeowners Report

13. Guardian: Premium Rises

14. MarshBerry: M&A Trends

15. IIAB: Consolidation

16. Salesforce: Agentic AI

17. Medium: Gen AI Role

18. McKinsey: Future of AI

19. McKinsey: Insurance 2030


Shannon Diem

Profile picture for user ShannonDiem

Shannon Diem

Shannon Diem is the founder and CEO of Intelligent Voice AI, the company behind Evolve AI Agents, agentic voice AI built for healthcare, insurance, legal and home services businesses. 

He is also the author of AI Unleashed! The CMO's Strategy for Smarter Marketing with AI

Insurance Fraud Targets Seniors Online

Scammers increasingly impersonate insurers to steal identities and money during open enrollment.

Insurance

Last year, the Justice Department brought the largest healthcare fraud case in its history, which included charges that criminals stole the identities of more than 1 million Americans to submit $10.6 billion in fraudulent Medicare claims for urinary catheters. While Medicare blocked nearly all of it, private Medicare supplemental insurers paid out about $900 million.

Americans lost a record $20.9 billion to internet crime in 2025, up 26% in one year, according to the FBI. Insurance fraud, including healthcare fraud, costs Americans more than $300 billion a year, and health insurance scams pick up during open enrollment, the FTC says.

Seniors are the hardest-hit age group, losing $7.7 billion in 2025, up 59% in one year. In fact, Michigan officials warned on Oct. 12 of impostors claiming to be from Medicare to dupe older adults.

As co-founder of Spokeo, a people search company, and former co-chair of the W3C Credentials Community Group, I have witnessed firsthand how scammers have grown increasingly sophisticated over the last 20 years.

Insurance scams are evolving and costly

We are seeing many insurance-related scams in 2026. Fraudsters are buying sponsored search ads with web addresses and logos that look very similar to HealthCare.gov and Medicare.gov. When Americans click on the link, they end up buying a plan that is not real insurance and also fork over their personal information, which can lead to identity theft.

Even large health insurers are being duped, as the $900 million loss in the catheter case shows.

The Coalition Against Insurance Fraud calls insurance fraud "the crime we all pay for," because it comes back to honest policyholders as higher premiums, which can hurt retention.

In a September survey by BrandShield, 30% of U.S. adults said they would stop buying online from brands scammers had impersonated. For an insurer, that is a renewal lost to a scam it never ran.

Here are three tips for insurers to help prevent insurance fraud
  • Say what you will never do. Tell customers you will never ask for a one-time code or take payment by gift card, wire or crypto. Put the statement on the ID card, in the app, and on every renewal notice.
  • Explain what numbers to trust and verify. Scammers can spoof caller ID so your company's name appears on a customer's phone. In every fraud alert, encourage customers to use the number on their card and check the identity of incoming callers with a reverse phone search.
  • Report fake ads and websites and warn consumers. Search regularly for impostors using your name and report them to the FBI IC3, the FTC, and the online platform where they live. Send alerts to customers in real time.

Harrison Tang

Profile picture for user HarrisonTang

Harrison Tang

Harrison Tang is co-founder and chief operating officer of Spokeo, a search service that has helped 350 million users avoid scams since 2006. 

He spent three years as co-chair of the W3C Credentials Community Group, which develops open standards for verifiable digital credentials.

The Claim Runs on the Data Captured First

Carriers investing in AI-driven claims automation face a data quality crisis rooted in fragmented, inconsistent loss documentation from the field.

Insurance

For years, loss documentation has been treated as a checklist. Get the photos, get a sketch, get the measurements, close the file. That worked when the file's only job was to support one estimate. It does not hold up now. The capture from the first visit has become the foundation the rest of the claim is built on, and increasingly the foundation that automation and artificial intelligence are built on, too.

So the question that matters is no longer whether the field got good photos. It is whether the capture produced data the rest of the claim can actually use. A scan with inconsistent measurements or missing coverage does not just put one claim at risk. It becomes a data quality problem that follows the file everywhere it goes.

Fragmented capture is a data problem

Most loss documentation is still assembled from separate tools. Photos on a phone. A sketch in one app. Measurements in another. Each piece is its own file, in its own format, captured in its own moment.

That fragmentation is where the data problems start. Gaps open between the pieces. The photos cover one set of rooms, the sketch another. A measurement in one tool does not match what the photos show. None of it was built to work together, so the mismatches compound as the file moves downstream, and a weak capture ends up weakening everything built on top of it.

Carrier claims leaders feel this even though they never pick up a camera. What they need from a file is data they can trust and the full context of the loss. A folder of photos gives them neither. It shows the angles someone chose to shoot, not the loss in context, and it leaves the person reviewing the file to fill the gaps from memory.

What claim-ready data actually requires

Claim-ready data is not better photos. It is complete coverage the field can confirm on site, so no one drives back for the room they missed. It is consistent, structured damage tags instead of loose notes. It is measurements the estimate can trust. And it is a record that drops into the estimating platform instead of being keyed in by hand.

Completeness is the part that pays off later. When the whole property is captured, affected and unaffected areas alike, the record answers the disputes that actually come back. A homeowner reports new damage in the next room weeks later. A capture of that room from day one shows whether the damage came from the original loss or from something unrelated. When the rebuild scope gets challenged, the file shows the full loss, room by room, not a reconstruction from memory.

That context exists for one moment only, before the cleanup starts. You cannot shoot it again after the fact. If it was not captured on the first visit, it is gone for good.

The data has to move

Complete, structured, trustworthy data is still worth nothing if it sits on a field device. In a contested or high-value loss, the people who need to see the scene are rarely the ones who captured it. The desk adjuster setting scope. The examiner reviewing the file. Counsel deciding in the first week whether there is a subrogation case worth pursuing.

Here is the gap almost no one has closed. The technology to capture a full virtual walkthrough on site and send it as a link before anyone leaves already exists. Used that way, the desk gets eyes on the loss the same day, approvals move faster, and a subrogation opportunity gets flagged while the evidence is still on the floor. That last one runs on a short clock. Subrogation is usually decided in the first days, and the physical evidence, a failed valve or an appliance, is often gone within hours of mitigation.

In practice, this almost never happens. The contractor does not send the link. The adjuster does not open it when they do. The record sits in the field while the window that matters runs out. The carriers that close that loop change what documentation is for. It stops being an archive to pull from later and becomes a decision they can make the same day.

Getting it right the first time

This is where the stakes climb. Carriers are putting real money into automation and artificial intelligence to triage, estimate, and audit claims. Those systems do not learn from the active file alone. They learn from the archive. A decade of thin, inconsistent photo sets is not a dataset. A decade of complete, structured records is.

That raises the cost of getting the first capture wrong, because there is only one. So the real test for capture technology is no longer whether it takes good pictures. It is whether the field can verify coverage on the spot, whether the output is structured data instead of loose files, whether the measurements can carry an estimate, whether the record hands off cleanly into the estimating platform, and whether the organization can secure it and still reach it years later.

A carrier cannot operate every camera in the field. But carriers set the standard for what a loss file has to contain, and they decide how fast that file has to move. Set the standard for the data the whole claim will run on, not for the photos that clear today's file, and the record holds up every time the claim asks it to.


Kayla Theriot

Profile picture for user KaylaTheriot

Kayla Theriot

Kayla Theriot is an account executive at iGUIDE, a property documentation system that combines imagery, measurement and floor plans from a single scan. 

She came to it from the field, with more than five years in restoration at a Servpro franchise in Louisiana and a role managing regional production at a fire O&C firm, and she holds the IICRC water damage restoration technician certification. 

Why Can't We Fund Prevention?

Industrial organizations excel at financing recovery from asset failures but struggle to justify prevention spending with the same economic rigor.

Insurance

We've built an extraordinarily sophisticated financial system around industrial failure. We can price it, insure it, reserve capital for it, and finance recovery when it happens, but preventing that same failure often requires an entirely different standard of proof.

Insurance is exceptionally good at mobilizing money after something goes wrong. Natural catastrophes caused an estimated $100 billion in economic losses in the first half of 2026. Insurance covered approximately 42%, leaving nearly $60 billion to be absorbed elsewhere. Insurance's ability to fund recovery is indispensable. The shortfall also exposes a contradiction in how risk is funded.

Premiums, deductibles, and recovery costs have established places in corporate budgets. A proposal to prevent an asset failure faces a tougher test. Someone must prove the threat is serious, determine what intervention it warrants, and defend spending money on an event that might never happen. Failure has a funding mechanism. Prevention has a business case to make.

Insurance can transfer part of a financial loss. It cannot restore missed production, repair damaged customer relationships, or undo the wider consequences of an industrial disruption. The challenge is making prevention as intelligible and investable as recovery.

Knowing the risk isn't the same as managing it

Industrial organizations rarely lack information. They receive risk-engineering reports, inspection findings, maintenance records and alerts from connected equipment. Yet information doesn't guarantee action.

A temperature change, unusual vibration, or a deteriorating component has little meaning in isolation. Its significance depends on the asset's role in the operation. Is there redundancy? What production depends on it? What would a shutdown cost? Could a failure disrupt customers, supply chains or other facilities?

These questions turn technical observations into business consequences. Without that translation, recommendations can remain in reports while maintenance teams address more immediate demands and capital committees fund projects with clearer returns.

Industry has invested heavily in detecting changing conditions and predicting potential failures. Those capabilities matter. Still, knowing what might fail doesn't tell an organization which emerging risks deserve attention first.

A relatively minor condition on a critical asset may represent far greater exposure than a severe condition on equipment with sufficient redundancy. The challenge is understanding what matters most, what the consequences could be, and where intervention will create the greatest value.

Prevention needs to compete for capital

Consider a manufacturer that identifies deterioration in a critical component on its main production line. Although the component hasn't failed, its condition is changing. The organization must compare the cost of replacing it during a planned shutdown with the lost output, repair time, and disruption that could follow an unplanned failure.

The consequences, however, may extend well beyond that production line. A shutdown could affect customer deliveries, contractual commitments, supply-chain continuity and financial performance. What appears to be a maintenance decision may represent a much larger business risk.

The appropriate response might be immediate repair, closer monitoring, a change in operating parameters or acceptance of the exposure. Someone must have the authority to make that decision and commit the necessary resources before the opportunity to intervene passes.

Now multiply that challenge across hundreds of assets and facilities, each presenting different levels of exposure. The question is no longer simply whether an individual intervention is justified. It's which interventions deserve priority when resources are limited, and the consequences extend across the enterprise.

For decades, industry has developed sophisticated ways to measure losses after they occur. We need comparable discipline in understanding the economic value of preventing those losses before they happen.

Capital should not only be deployed to absorb loss. It should be possible to direct it intelligently toward preventing loss.

The standard for resilience must rise

Risk assessments, inspections and engineering recommendations are essential. Completing those processes, however, doesn't necessarily mean an organization has addressed its most consequential exposures. An inspection can establish that a process was followed. A risk assessment can document a vulnerability. Neither, by itself, demonstrates that the organization is becoming more resilient.

Industrial risk isn't static. Asset conditions change, operational dependencies evolve, and an exposure that appeared manageable during a periodic assessment may become significantly more consequential.

A recommendation without a named decision-maker and deadline is information, not control. Someone must have the authority to accept the exposure, fund the response or explain why another course is justified. Following the process is important, although it's not the same as preventing the loss.

Insurance can help change the economics of prevention

Insurers can make recommendations easier to act on by explaining the severity and urgency of an exposure, not simply recording that it exists. Brokers can connect risk-engineering findings with an organization's insurance and resilience priorities. Operators must determine whether to intervene or accept the exposure and take responsibility for the outcome. There is also a bigger opportunity here.

Insurers and brokers possess extraordinary expertise in understanding, quantifying, and pricing risk. Much of that expertise has historically helped organizations prepare financially for potential losses. What if more of it could be applied to making prevention economically compelling before those losses occur?

Imagine insurers, brokers, and industrial operators working from a more continuous, consequence-based understanding of emerging physical risk. Rather than relying primarily on historical losses and periodic assessments, they could better understand where exposure is developing, which conditions present the greatest business consequences, and where intervention could materially reduce risk.

This wouldn't eliminate uncertainty or diminish the need for insurance. It could, however, create a stronger economic foundation for prevention and a more productive relationship between risk financing and operational decision-making.

For decades, the insurance industry has helped businesses understand the financial consequences of what can go wrong. That expertise will remain indispensable, but the next opportunity is to help make the economics of what doesn't have to go wrong equally visible.

We've become extraordinarily good at financing industrial failure. It's time to become just as disciplined about investing in its prevention.


Andy Pruett

Profile picture for user AndyPruett

Andy Pruett

Andy Pruett is founder and CEO of Lumicent.

He began his career in the US oil and gas industry, working across upstream, midstream and downstream operations, before moving into mining and helping build the Pogo gold mine in Alaska. Today, his work centers on decision intelligence for physical operations: connecting changing asset conditions with their operational and financial consequences so organizations can prioritize risk, direct resources effectively and act before preventable exposure becomes a loss.

Why Insurers' Internal Software Builds Never Finish

Internal software builds at insurers routinely stretch past two years and $30 million because of a structural misunderstanding. 

Insurance

This year I spoke with several people in insurance and insurtech whose internal software builds started in 2025. Almost two years later, they are still building. The spending they described ran from $12 million to $30 million plus. None of these teams lacked smart people. The problem is structural, and it is worth naming so carriers and MGAs can plan around it.

Core versus context

Geoffrey Moore's distinction between core and context is a useful lens. Core is what the market pays you for. Context is everything else that has to work but does not differentiate you. For an insurer, core is underwriting, claims, and distribution. Software is context.

At a software company, the engineer is the product, and the budget, the career ladder, and the roadmap all follow from that. At an insurer, engineering is a support function. It competes for budget with claims, underwriting and compliance, and its roadmap gets bumped by the next regulatory filing or system outage. None of that reflects on the engineers. It reflects where engineering sits in the business, and the strongest engineers tend to go where it is the product.

What the data shows

The numbers point the same way, with caveats on scope.

MIT's NANDA initiative studied generative AI deployments across industries, drawing on 150 leader interviews, a 350-person employee survey, and 300 public deployments. It found that purchased tools and partnerships succeeded about 67% of the time, and internal builds succeeded one-third as often. That study covers AI projects broadly, not insurance alone, and its headline failure rate has been debated, but the build-versus-buy gap is consistent with what insurers report about their own technology.

A West Monroe survey of 300 insurance executives, covered here on ITL in December 2025, found that more than half of insurers spend 51% to 75% of IT budgets keeping existing systems running. Two-thirds expect another three to seven years just to move core systems to the cloud. And 52% said they had delayed or canceled two to three strategic technology programs in the past 12 months, citing budget constraints.

An internal team inherits that environment. The people who should be building the new system are often the same people holding the old one up.

The coordination problem

Fred Brooks explained a second cause in The Mythical Man-Month. The number of communication paths in a team of n people is n(n-1)/2. A team of six has 15 paths. A team of 60 has 1,770. Ten times the people means 118 times the coordination. Adding headcount to a stalled build usually adds more overhead than output.

What to do instead

None of this means insurers should never build. It means they should be deliberate about what they build and who builds it.

First, buy where a mature product exists. The MIT data suggests purchased tools and partnerships succeed more often, so reserve internal builds for what truly differentiates the company.

Second, when something has to be built, keep the team small and senior. Six people who have shipped products quickly, who understand how carriers and MGAs actually operate, and who are accountable for the result will usually move faster than 60 people spread across committees.

Third, scope to one workflow and set a go-live date. A named workflow, real policy and claim files from day one, and security and compliance involved before the first build, remove most of the surprises that stall projects.

Fourth, measure the project by production use, not by activity. A system that adjusters and underwriters do not use is not finished, however much has been spent.

The question for leaders

If your longest internal software project has been running for more than a year, the useful question is not who is to blame. It is whether the structure around that project, including budget priority, team size, and ownership, can ever produce a finish. For many insurers, the honest answer is no, and the fix is a structural one.


Anderson Whittle

Profile picture for user AndersonWhittle

Anderson Whittle

Anderson Whittle is CEO of WYEA, an engineering firm that builds AI workflows into the policy and claims operations of specialty insurers, MGAs, program administrators and reinsurers.

November 2026 ITL FOCUS: Pricing & Rating

ITL FOCUS is a monthly initiative featuring topics related to innovation in risk management and insurance.

 
 
 
 
 
November 2026 - Pricing & Rating.png

FROM THE EDITOR

Pricing is one of the most consequential functions in insurance — and one of the least visible. Claims get the headlines. Underwriting gets the scrutiny. But pricing sits quietly underneath both, shaping profitability across millions of policies through assumptions that can go stale faster than anyone realizes.

For most carriers, the mechanics of pricing haven’t kept pace with how quickly the world is changing. A rate change approved today might not be live in the market for weeks or months, trapped behind policy administration release cycles that were never designed for the kind of volatility the industry is now dealing with — social inflation, rising repair costs, tariff pressure, and catastrophe losses that keep defying historical patterns.

Kumarappan Sundaram has spent years working through these problems with P&C carriers as a Managing Director at PwC. He brings a technology practitioner’s perspective to a function that often gets treated as a back-office infrastructure problem rather than a strategic capability. He’s measured about where AI fits — distinguishing what is already delivering clear value from what remains genuinely unproven — and direct about the sequencing mistakes carriers make when they try to modernize pricing before the foundation is in place.

His practical advice is worth taking seriously: Before reaching for advanced analytics or AI, measure how long it actually takes to get an approved rate change into the market. Most carriers have never done that math. The number, he suggests, tends to be illuminating.

Read the full interview to find out why decoupling rating logic from the core policy system is the starting point most carriers skip, how modern pricing platforms are beginning to close the long-standing gap between pricing and underwriting, and where AI can genuinely change the work.

 

 

 
November 2026 Kumarappan Sundaram .png

Pricing at the Speed of the Market

Paul Carroll, Editor-in-chief of Insurance Thought Leadership.

Pricing gets far less attention than claims or underwriting, yet it is important. How is the pricing function being modernized in insurance? And how should it be?


Kumarappan Sundaram, Managing Director, PwC US

Pricing gets less attention partly because a lot of the work is infrastructure: platforms, data flows and deployment processes. It is not a front-office workflow that a new tool can visibly transform.

Modernization can be uneven. Personal lines are well on their way. Commercial lines are earlier in the journey and still rely heavily on underwriter judgment and rating engines embedded in policy administration systems.

How pricing should be modernized can largely be a question of sequence. The foundation comes first: decoupling rating from the policy admin system. At the moment, actuaries produce pricing logic, which they hand to IT, which embeds the logic in a rating engine that sits within a policy admin system. So, making a change to the pricing logic means waiting for a new release of the policy admin system. If you decouple, you can make changes to ratings as and when you want.

Once you’ve decoupled, you should build reliable data flows and establish clear governance.

You can then go to the next level of capabilities, like advanced analytics and AI. Let’s say you have a $10 billion book. You’re considering changing a particular rate factor from X to Y, maybe related to driver age. Advanced analytics can help you understand the impact of the rate change on the whole book. The actuaries can then decide whether to make that rate change. Carriers can help deploy and monitor pricing changes quickly and safely.


Paul Carroll

From a technology standpoint, what are the most important shifts in P&C pricing and rating?
 

Read the full interview >

 

MORE ON PRICING & RATING

pexels-iriser-1499269.jpg

Continuous Underwriting Wants to Scale 
 

by Riv Arthur

Insurance premiums could fluctuate daily like stock prices, but regulation and reinsurance prevent the scaling of continuous underwriting.
 

Read More

itl & six things hero image - 2026-09-16T150722.149 (4).png

Agentic AI Redefines Real-Time Underwriting

by Rachita Nahata

Real-time underwriting isn't about instant decisions but leveraging agentic AI and current data to help underwriters assess risk faster.
 

Read More

Six things 08.09.22 (1) (2).png

Breaking Down Silos in Insurance

by Rory Meleniclis

Insurers can dismantle organizational silos by shifting from project-based work to product teams built around shared capabilities like underwriting and claims.
 

Read More

itl & six things hero image - 2026-10-12T131112.611 (3).png

How AI Can Transform Submissions for Underwriting

by Prathap Gokul

Commercial insurers face a strategic challenge: identifying which submissions merit scarce underwriting attention, not just processing more faster.
 

Read More

itl & six things hero image - 2026-07-23T102617.356 (2).png

Making AI Work in Commercial Submission Intake

by Vital Soupel

Insurers need a practical blueprint for where to apply AI first, how to move beyond pilots, and how to scale the technology responsibly.
 

Read More

An Interview with Samit Bhandari, PwC (4).png

Turning Volatility Into Value

by Insurance Thought Leadership

Amid today's heightened uncertainty, Dawn Miller, chief commercial officer at Lloyd's, explores how insurers can innovate their way to value both for themselves and for their clients. 
 

Read More

 

MORE FROM OUR SPONSOR

AI adoption in P&C insurance: From token cost to business-unit value

Sponsored by PwC

See how a human-centered approach to change helps your people adopt new ways of working—so transformation sticks.

Read More

 


Insurance Thought Leadership

Profile picture for user Insurance Thought Leadership

Insurance Thought Leadership

Insurance Thought Leadership (ITL) delivers engaging, informative articles from our global network of thought leaders and decision makers. Their insights are transforming the insurance and risk management marketplace through knowledge sharing, big ideas on a wide variety of topics, and lessons learned through real-life applications of innovative technology.

We also connect our network of authors and readers in ways that help them uncover opportunities and that lead to innovation and strategic advantage.

Smarter Approach to Prior Authorization

Healthcare plans are slashing prior authorization requirements, but smarter technology and mutual accountability may improve care more than blanket reductions.

Insurance

Prior authorization has been a healthcare sticking point for years.

Patients and providers criticize prior authorization (PA) for denying care and coverage. State and federal lawmakers target it, setting standards and policies for how health plans must operate, often without a deep understanding of the complexities of the issue. In response to the public and legislative pressure, many plans are reducing their reliance on PA.

UnitedHealthcare announced that it will cut PA requirements by 30% by the end of 2026, including some outpatient surgeries and diagnostic tests, as well as certain outpatient therapies and chiropractic care.

UnitedHealthcare's announcement is just the latest in a series of moves by plans to reduce their reliance on PA. In June 2025, industry organization AHIP committed to six PA reforms across commercial, Medicare Advantage and Medicaid managed care coverage, including reduced requirements and faster response times. Individual plans also have acted. AHIP reported earlier this year that major insurers had cut 11% of PAs (6.5 million).

While reducing PA will certainly eliminate some friction, and please patients and providers, blanket policies are seldom the solution to an issue this complex. Rather, this is an opportunity to take a smarter approach to PA, one that improves care and makes it smarter, fairer, and more efficient for all involved. One that treats PA as a tool for better care rather than a barrier.

The cost of poor prior authorization

There is no doubt that PA needs to improve. A process that started in the 1960s with the launch of Medicare has become far too cumbersome. Providers say it delays or prevents treatment, adds administrative costs, and overrides their clinical judgment. A 2026 AMA survey found an average of 40 requests per physician per week, taking 13 hours of physician and staff time.

Members become frustrated with their plans when approval isn't granted, often without explanation.

Plans counter that PA helps ensure appropriate care, block unnecessary or unsafe care, and control costs.

Despite all the heat surrounding this issue, there is room – and need – for compromise. Technology has made it possible for PA to be handled far more quickly than in the days of fax machines and phone calls. Both plans and providers also have access to science-backed policies that can better inform their decisions.

Rethinking prior authorization

It's important to understand that the problems with PA are not the fault of any single party. There needs to be improvement and mutual accountability on both sides in order to improve the process.

Plans owe providers and their members transparency in their policies and decision-making, consistent turnaround on claims, and fast, clear decisions. Providers owe plans accurate, evidence-based ordering and complete documentation. This is a precondition for any PA reform to actually work. Without mutual accountability, reducing mandatory PA checks just shifts risk around instead of removing it.

It's time to reset our idea of PA from a barrier to care or an indiscriminate way to control spending, to a tool for ensuring patients receive the right care at the right time for the right reason. It should guide and clear the way for timely, effective care.

Lab Benefit Management

An example of this can be found in lab benefit management (LBM), the governance of lab testing. Clinical testing is the most common medical procedure. Approximately 14 billion lab tests are performed annually, and the results inform 70% of healthcare decisions. However, because individual tests are generally inexpensive, they are not as scrutinized as other procedures. The result is that as much as 30% of testing is unnecessary or redundant, wasting billions of dollars and hindering care.

LBM tackles this problem by creating a science-backed set of policies for individual plans. It identifies the tests that offer clinical utility and validity and uses those findings to inform coverage decisions from plans and guide clinicians toward the most effective treatments.

However, creating accurate, science-based policies is only the first part of a comprehensive LBM program; the other half is implementing it. Much of the problem with PA is that it is reactive: a clinician submits a request; the plan may reject it for any number of reasons, and it comes back to a frustrated clinician and member, who must try to resolve the issue.

Here's a better way: An automated solution trained in plan-specific policies and coverage is integrated into providers' EHRs. It spots inaccurate or missing information in a PA request and flags it for the provider to correct before submission. Once the request is in order and meets the necessary criteria, the solution automatically approves it, removing uncertainty. It also lets the provider know whether the test or treatment is not covered for that patient or if the plan prefers an alternative. A provider can then decide whether to choose another, more appropriate test or proceed with further review by the health plan. In essence, the new process moves the appropriateness review farther upstream to the moment the provider enters an order for a test or procedure, saving time and resources.

There are other examples of innovation that link outcomes and PA. The new CMS Cell and Gene Therapy (CGT) Access Model, built around sickle cell disease gene therapies, involves government and drug manufacturers as well as plans and providers. CMS negotiates outcomes-based agreements directly with gene therapy manufacturers on behalf of states. The manufacturers provide guaranteed discounts and pay rebates back if the therapy doesn't deliver the promised clinical benefit. This shifts some of the financial risk from states to manufacturers.

This model works because of a safeguard baked into its design: strict upfront criteria to confirm which patients are truly eligible for the therapy. For a one-time, irreversible treatment costing millions of dollars, that kind of appropriateness review isn't a barrier to access — it's what makes access sustainable. The outcome-based rebates manufacturers and states are counting on only hold up if the therapy reaches the patients it's meant for. In that sense, the CGT model and smarter prior authorization share the same logic: verify appropriateness upfront, and both access and affordability take care of themselves.

The case for smarter prior authorization

The need for a better process to ensure high-quality, appropriate, and affordable care is made more urgent by the explosive growth of precision diagnostics and genetic medicine. These breakthroughs are transforming medicine, offering unprecedented accuracy in diagnostics and new treatments for diseases once thought incurable, but they are expensive and complex.

A process like the one described above will remove providers' doubt about which tests are appropriate, streamline the approval process, and enable patients to receive treatment faster.

As healthcare organizations and legislators consider PA reform, they should ask themselves if their plans only reduce risk and friction, or whether they truly improve the quality of the care that is being delivered to patients while insisting on accountability from all parties.

How Gen AI Can Help Shape Strategy

At any level of setting strategy, from the boardroom to the one-person agency, generative AI can bring assumptions to the surface, helping you see what NOT to do. 

Image
Insurance

Bill Gates recently said he is using generative AI as part of the annual, two-week strategy session that determines how the Gates Foundation spends its next $10 billion. He'll be "taking the strategy notes and actually engaging in a dialogue with ChatGPT, Claude, Copilot, and even having them talk with each other," he said.

I don't have $10 billion that I'm considering how to spend, and you may not, either, but genAI has now improved to the point where it can at least be an aid in setting strategy at almost any level. 

An old boss, Mel Bergstein, used to say: "Strategy is what you don't do." The idea worked for Mel, who founded a consulting firm, Diamond Management & Technology Partners, that pioneered the notion of digital strategy in the mid-1990s. Chunka Mui and I took Mel's adage further in our book, "Billion Dollar Lessons," for which we had 20 researchers spend two years looking at 2,500 major failures to identify the warning signs.

Our recommended solution boiled down to having a "devil's advocate" identify all the underlying assumptions for a strategy, to keep problems from being glossed over as an executive team rushed to realize the CEO's plan. GenAI, used right, can now do a pretty good job of filling that role for anyone.

Let's have a look. 

Gates told the New York Times that, "in a few of the reviews, we’ll actually have the A.I. sit in. In a few cases, we’ll tell it: Hey, only speak if we ask you. And then in a few other cases, we’ll say: Hey, if you hear something you think is wrong or you hear us thinking about, for example, what are these statistics? Please engage. So we’ve gone from it being a joke [when someone suggested using genAI a year ago] to it will be a peer — not making any final decisions, but it will be a peer in deep, complex strategic discussions, making a significant contribution."

I'm not willing to go that far just yet, partly because I'd need to test the domain knowledge that the large language models have about a subtle, heavily regulated business like insurance and partly because genAI has shown itself to be a sycophant. The biggest, overarching pattern of failure that Chunka and I found in the research for "Billion Dollar Lessons" was that people were reluctant to point out obvious problems that could derail initiatives that had the imprimatur of the CEO, and we don't want to replace human sycophants with AI ones. 

But you can describe your strategy to an AI, then pepper it with questions, going deeper each time while demanding evidence for its thinking and insisting it be blunt.

Let's say you're thinking of buying a company. You could start with a general question: What could go wrong? Then you probe. How will our competitors react? What if interest rates increase? The economy goes into recession? The war in Iran escalates? Will we actually be able to cross-sell to our customers? Will the efficiencies from scale be as great as we expect? Who's done something similar and had problems? What were the problems? And on and on and on. 

You can do the same exercise if you're thinking of a new line of business, a software overhaul, adopting AI chatbots — or if you want to explore the dangers of standing pat, of not undertaking a major initiative that is under consideration. 

Major corporations have staff do extensive analysis along the lines I'm suggesting and, in the case of a possible acquisition, draw on their investment banks for advice — but the bias is always toward acting. If the perception is that the CEO backs an idea, the staff strives to make it work, not to kill it. Investment banks only collect their hefty fees if the deal happens, so why would they recommend against it? 

When Chunka and I did a consulting project with a major company that was considering launching autonomous vehicles, we saw elaborate studies it had commissioned from consulting firms that said the robotaxi business was a slam dunk and that the first to enter would win big. That was more than eight years ago. 

It turned out, however, that the executives working on the AV project internally were aware of all the sorts of operational and safety issues that we've all seen play out and that have slowed the robotaxi rollout. Those concerns were brought to the surface in the interviews for our devil's advocate review. 

Warren Buffett used to set aside two, equal pots of money when asking investment banks for advice on possible acquisitions — one for a bank that argued in favor of a deal and one for a bank that argued against it. Whichever carried the day collected its fees, while the other got nothing. You can be sure the "anti" bank worked as hard as possible on its arguments and evidence. 

In my experience, many big ideas should fall apart at the slightest scrutiny. As we wrote in "Billion Dollar Lessons," we thought 47% of the failures we investigated were based on strategies that never should have made it out of the strategy room — and not because those involved were stupid or inexperienced. 

The fifth-biggest cement maker in the world reasoned that it was really in the home products business because so much of its cement was used there. It began also making furnaces and lawn mowers... then went bankrupt and was bought by a competitor. Avon decided that its core competency was its culture of caring — not, you know, its line of cosmetics and its devoted salesforce — so it bought a maker of medical equipment and an operator of retirement homes. It had no idea how to operate the new businesses and quickly sold them at a major loss. In the early days of fax machines, FedEx spent hundreds of millions of dollars to set up a business that would collect a document from you, drive it to the nearest office, fax it to the office nearest your destination, then drive the fax to the recipient. The company somehow missed that Moore's law would soon make the devices so inexpensive that every company would have its own.

It may take time to learn how to use genAI to avoid costly strategic mistakes — and that's even before we start using it as broadly as Gates is planning to. You'll likely also need to introduce to the LLM information specific to your market, including regulatory issues that it likely doesn't have even as these models try to Hoover up everything available on the internet. Insurance is a complicated market.

But I think we're now at the point where genAI can, if managed correctly, start to catch strategic errors. We won't get that 47% figure for obviously bad ideas down to zero, but even heading toward 40%, maybe even 33%, would mark real progress.

Cheers,

Paul