InsurTech: prove necessity

A lot of insurance technology companies spent the last few years selling “artificial intelligence for insurance,” and investors are now asking a simpler question: does the software get installed, used, and renewed by a real carrier. CB Insights said insurtech dealmaking in 2025 shifted toward later-stage companies with measurable traction, and all 11 startups that raised $100 million or more had already reached its “scaling” category. (cbinsights.com) Insurance runs on old core systems that handle policies, billing, and claims, and many of those systems were built long before modern application programming interfaces existed. A 2025 Geneva Papers review said legacy information technology systems are still a major obstacle to agility, which is why startups that plug into existing workflows have an easier sell than startups asking carriers to rebuild the whole house. (link.springer.com) That is why “integration” has become a business test instead of a technical detail. The insurance technology group IB Applications argues that connecting quote, bind, policy, and claims tools into existing carrier systems cuts manual rekeying, speeds processing, and removes the kind of workflow friction that makes staff abandon new software. (ibapplications.com) The money side of insurance makes this harsher than in many other software markets. Verisk and the American Property Casualty Insurance Association said the United States property and casualty industry posted a 96.5 combined ratio in the first half of 2025, down from 97.6 a year earlier, and that ratio is watched closely because every point can mean billions of dollars in underwriting profit or loss. (verisk.com) So an insurtech pitch now has to land in one of two places: help underwriters price risk more accurately, or help operations teams do the same work with fewer errors and fewer people touching the file. CB Insights said future comparisons in insurtech are likely to center on measurable performance like underwriting results, growth, and delivery at scale rather than pilot outcomes. (cbinsights.com) That is also why data quality and governance are suddenly getting more attention than flashy models. If an insurer’s policy records, claims history, and external risk data are inconsistent, the model sits on top like a calculator with the wrong numbers punched in, and the output is still wrong even if the interface looks futuristic. (link.springer.com) (ibapplications.com) Trade coverage in Ireland captured the mood shift clearly this week. Silicon Republic quoted InsTech.ie chief executive Gary Leyden saying the Irish market is moving from “talking about innovation to proving it in practice,” and he pointed to firms like Inaza, Blink Parametric, and Docosoft as examples of companies already improving underwriting, real-time products, or regulatory workflows. (siliconrepublic.com) The phrase carriers use for the opposite problem is “proof-of-concept purgatory.” Leyden described a pattern where startups can demo tools but struggle to test them in live insurance environments, while insurers keep extending pilots because operational risk, compliance, and system complexity make full deployment harder than the sales deck suggests. (siliconrepublic.com) That leaves the strongest insurtechs looking less like moonshot disruptors and more like specialist contractors replacing bad plumbing inside a very expensive building. The Geneva Papers review said the digital overhaul of core systems is what creates the foundation for new business models, which means the boring work of data pipes, controls, and process design now sits upstream of the exciting work of machine learning. (link.springer.com) Investors are still funding the sector, but the bar has moved from “could this transform insurance” to “show me the loss ratio, the renewal rate, and the integration timeline.” In 2025, only 24 percent of insurtech deals tracked by CB Insights went to startups in its earliest commercial-maturity categories, a sign that capital is following proof over promise. (cbinsights.com)