Carriers Need to Embrace Machine Learning, Says Former Litigator Turned InsurTech Exec

This article is part of Carrier Management’s series on the Future of Insurance.

Adam Cassady, CEO, Tyche Risk, believes insurers that embrace machine learning technology to garner better risk insights will be better positioned than those that cling to their beliefs in the “art” of underwriting in the future.

Adam Cassady, CEO, Tyche Risk

Adam Cassady, Co-Founder and CEO, Tyche Risk https://www.tycherisk.co/

Through an interactive platform, Tyche delivers undiscovered information to commercial casualty underwriters to help them make smarter decisions. By blending open data and a client’s own proprietary data with machine learning, Tyche builds predictive claims-avoidance models that can help clients to shrink the riskiest fractions of their books of business.

Cassady is a former insurance coverage litigator who says that lawsuits over health issues linked to the microwave popcorn flavoring chemical Diacetyl, and the corresponding insurance claims, sparked the idea for a model that would provide better risk insights to underwriters.

Q: What major changes do you see on the horizon for the property/casualty insurance industry in the next 10 years?

Cassady (Tyche): Machine learning technology and its proponents will butt heads with hard-won but increasingly anachronistic insurance industry orthodoxy that the core competencies of underwriting and claims management are art and are not amenable to fundamental transformation through data science. While each exposure and each claim is unique, the laws—both natural and human—governing those exposures and claims are increasingly both knowable and known. Machine learning technology enables us to discover, describe and deploy knowledge of those laws, and do so quickly.

“Carriers can avoid costly and pointless disputes and needless discovery by communicating claims settlement offers that are north of what the carrier would ultimately have settled at, delighting and doing right by their policyholders while still saving money on loss adjustment.“

In the near future, humans will still be steering the ship of underwriting and claims management. But those carriers that, in 10 years, have not fundamentally rethought how those humans function in relation to already deeply deployed machine learning technology—and I mean specifically machine learning, not “automation,” or “digital,” or “big data,” whatever those terms mean—will be at a significant disadvantage compared to competitors. The size of the disadvantage will be a function of how successfully cultural friction in the industry slows the integration of machine learning technology.

Those who leap ahead in this area will have bigger books because they will be able to quote submissions much more quickly and triage to improve hit ratios. Those books will have better loss ratio characteristics because risk segmentation will be far better. Claims generated from those books will have lower loss adjustment expenses, because carriers can avoid costly and pointless disputes and needless discovery by communicating claims settlement offers that are north of what the carrier would ultimately have settled at, delighting and doing right by their policyholders while still saving money on loss adjustment. And overall expense ratios will creep downward as the productivity of new hires, enabled by machine learning technology, dramatically improves.

Q: What will insurance companies, insurance leaders, the industry and its workforce look like in the next decade?

Cassady (Tyche): In the next decade, the insurance industry and its leaders will look basically the same but will be poised for true transformation in the next 10 years [the 10-year period starting 10 years from now]. Nevertheless, a few trends are clear.

First, there will be fewer insurance agents and brokers because the workforce is aging out. But those that remain will be more productive, enabled by technology.

Second, InsurTech online/direct distribution platforms will evolve—first toward MGAs with the pen, and then toward fronting carriers working directly with reinsurers. We will see many more such entities in 10 years.

Third, carriers will continue to shave staff and partner with delegated authorities and TPAs, but the power and influence of the CIO/CTO will, rightly, grow. That does not necessarily mean their budgets will grow (nor should it), but technologists will inevitably begin to influence strategy. Insurance CEOs who are not asking their CIOs how they can better serve their customers, and closely listening to their answers, are doing themselves a disservice.

Fourth, both carriers and reinsurers will turn toward adding value beyond providing an insurance product. Carriers will start to deploy their underwriting knowledge, and more importantly data, toward proactive engagement with their insureds through loss prevention. Reinsurers will start to, either independently or with brokers, work harder to be a greater technical partner to their cedent clients.

Lastly, we’ll see some really interesting things happening on blockchain. If I were to wager, I would bet that a walled garden blockchain platform aimed at disintermediating the reinsurance broker and permitting nontraditional investors like family offices to participate in ordinary reinsurance treaties will crop up. Smart contracts and reinsurance go together like peanut butter and jelly. If this doesn’t exist in 10 years, I will be shocked, and if it doesn’t exist in three years, albeit with limited adoption, I will be mildly surprised.

Q: What risks will they insure?

Cassady (Tyche): I actually don’t think that that many more domains of insurable risk will be broached. So basically, whatever gets insured now. I, of course, hope Trōv and others prove me wrong.

Q: How will insurance products and services be distributed?

Cassady (Tyche): Online and through agents. The overall share of direct-to-consumer distribution will grow.

Read more Future Insights by person