Earnix report says insurers must turn AI into business value

Earnix’s latest insurance trends report draws a clear line between AI that looks impressive in demos and AI that changes how insurers actually price, underwrite, service, and retain business. The message is blunt: adoption is broad, but operational maturity is still uneven, and that gap is where software buying priorities are shifting.

AI is everywhere, but scale is the real test

The 2026 edition, titled *The Race to Reinvent*, is Earnix’s fourth annual insurance trends report and is based on a global survey of 400 insurance executives from Australasia, Europe, the United Kingdom, the United States, and Canada. The headline finding is that 81% of executives say AI is embedded across most or some of their workflows, while 43% say it is integrated across most functions. At the same time, 80% are already experimenting with or planning to adopt generative AI within the next two years, which shows how fast the market is moving from curiosity to active deployment.

But Earnix is just as focused on the limits. Fifty-six percent of respondents favor a gradual path that keeps human intervention in place for at least the next three years, a sign that insurers still see AI as a supervised capability, not an autonomous one. That caution matters in P&C software because the stakes are not abstract: pricing accuracy, underwriting consistency, claims leakage, and customer treatment all sit inside regulated workflows where bad automation can create more problems than it solves.

Where the maturity gap is still widest

The strongest evidence that the industry is still early in its AI journey comes from the use cases that matter most operationally. Earnix says fewer than a quarter of insurers currently use AI for claims processing, at 23%, only 18% use it for policy issuance, and just 15% use it to predict churn. Those are not edge cases. Claims, issuance, and retention are core business processes, which means many carriers are still testing AI around the margins rather than embedding it where it can influence revenue and loss performance.

That gap has direct implications for product, pricing, underwriting, and governance teams. Product groups need AI that can support faster tailoring without breaking filing discipline. Pricing teams need models that can ingest timely data without losing explainability. Underwriting teams need decision support that improves segmentation and referral logic rather than replacing judgment with opacity. Governance teams, meanwhile, need controls that preserve compliance without freezing change, because the more personalized the customer logic becomes, the more scrutiny it attracts.

Earnix’s brochure version of the report makes that tension explicit by framing the study around where AI is delivering value, what is holding organizations back, and how data quality, legacy systems, regulatory pressure, and customer-centric strategies are becoming essential to sustained growth. That combination is exactly why AI buying in insurance is no longer about feature checklists alone. The vendors that win will be the ones that can prove their tools fit into approval flows, data architectures, and audit trails already used by regulated carriers.

Personalization is the prize, but governance is the price of entry

Earnix’s emphasis on personalization and governance gets to the heart of the software buying conversation in P&C. Personalization promises more relevant pricing, product design, and customer engagement, but it depends on cleaner and more connected data than many legacy stacks can deliver. Governance is the other half of the equation: every step toward more tailored experiences increases the need for documentation, testing, controls, and human oversight.

That is why the report’s framing is useful for buyers. Insurers are not simply asking whether a tool uses machine learning or generative AI. They are asking whether it can support repeatable decisions, whether it can be monitored, and whether it can survive regulatory review without forcing teams to rebuild processes every time a rule changes. In practice, the buying priority shifts from isolated automation to decisioning architecture, the layer that connects models, data, approvals, and customer actions.

The report also suggests that market pressure is pushing insurers toward measurable business outcomes. AI will be judged less by novelty and more by whether it improves pricing precision, product relevance, operational consistency, and the speed at which teams can respond to changing conditions. That is a much stricter standard, and it places software vendors under pressure to show how their platforms work inside real operating models, not just in pilots.

The 2024 survey shows how fast the bar has moved

Earnix’s current findings make even more sense when compared with its 2024 insurance trends survey, which covered 431 global insurance executives and was conducted with Market Strategy Group, LLC. That earlier study found 70% expected to deploy AI models using real-time data within two years, a sign that the industry was already looking beyond static analytics and toward live decisioning. It also exposed serious process friction: 51% said their company had paid a fine or issued refunds due to errors in the last year, 58% said rule changes take more than five months to implement, and 21% said they take longer than seven months. Nearly half, 49%, said they were behind schedule on modernization.

That is the backdrop for why the 2026 report feels less like a trend recap and more like a buying roadmap. Robin Gilthorpe, in the earlier research, said the industry was moving from discussion to action, and the latest numbers show that action is now being judged by execution discipline. Earnix’s leadership voice, including Kathy Klingler and Gilthorpe, is effectively arguing that AI is no longer the differentiator by itself. The differentiator is whether an insurer can operationalize AI without breaking governance, slowing compliance, or overwhelming legacy infrastructure.

What carriers and vendors should take from the new baseline

For insurers, the practical takeaway is simple: AI strategy now lives or dies in the gap between experimentation and repeatable production. For vendors, the report raises the bar in a different way. It is not enough to promise smarter models. The product has to fit the way insurers actually work, with controls for approval, auditability, human review, and fast adaptation when rules or market conditions change.

That is why the report matters beyond its survey numbers. It captures a market that has largely accepted AI as table stakes, but has not yet solved the harder problems of scale, governance, and business value. The next buying cycle will reward platforms that help insurers turn personalization into compliance-friendly precision, and turn AI from a headline feature into a dependable operating capability.