For all the noise about generative AI, most insurers are still in the early stages of figuring out what it actually means for underwriting, operations, and customer outcomes. Pilots proliferate, chatbots appear on websites, and proof‑of‑concepts are showcased at conferences, but many leaders are still wrestling with basic questions about risk, governance, and return on investment.
Two industry voices – one from a digital insurer and one from an infrastructure provider – suggest that the divide in the next phase of AI adoption may have less to do with eye-catching demonstrations and more to do with the unglamorous work of culture, plumbing, and process.
Jeff McKay (pictured right), CTO and co-founder of digital insurance provider PolicyMe, is blunt about how he sees the technology.
“AI is one of those incredibly transformational technologies. For us, we’re leaps and bounds faster than we were before AI – it’s a whole paradigm shift in what’s possible,” he said.
For McKay, (if done right) that “paradigm shift” can help smaller and digital players compete with larger incumbents.
“It’s a speed game now. The winners will be the ones who embrace AI fully and make it part of their DNA,” he said. “We’re not nearly as big as the large insurers, so it’s critical that we punch above our weight. AI is the tool that allows smaller teams deliver far greater impact.”
So far, many insurers have concentrated AI efforts on relatively contained problems: routing customer queries, assisting agents with scripts and suggestions, or offering straight-through decisions on very clean risks with narrowly defined rules.
McKay believes the next wave will push further into the messy middle of underwriting, where cycle times are long, and information is fragmented.
“Instant decisions are not yet realistic for every case - only the straightforward ones. The real breakthrough will be using AI to dramatically accelerate the complex decisions,” he said.
One area ripe for change, especially in life and health, is the way underwriters handle attending physician statements (APS).
“An attending physician's statement can be dozens, even a hundred pages. That’s incredibly labour-intensive to read. AI can summarize that in a fraction of the time and help underwriters move much quicker,” McKay said.
Used in this way, AI is framed less as a replacement for underwriters and more as a tool to strip away low‑value work.
Underwriters still decide; AI just gets the file to them faster and in a more digestible form.
“Ultimately, AI should be an accelerator for the customer, not just the company,” McKay added. “There are still long applications in insurance. The question is: can we tap into different data sets or let people upload documents so they don’t have to slog through a 15-minute form?”
Many carriers still treat AI as something that lives with IT or in an innovation lab, far from the day-to-day work of underwriters, claims handlers and operations teams, McKay said. He argues that, without broader cultural change, that approach will hit a ceiling.
“For us, AI is a first-class citizen,” he said.
That positioning, though, needs to be intentional. McKay stressed that giving staff permission and guardrails to experiment is as important as the technology choices themselves.
“We’ve built the infrastructure so people can use AI safely, then we push hard on education and training to demystify it,” he said. “We run internal hackathons and share real use cases to encourage its usage across the organization.”
He acknowledges that not everyone is enthusiastic.
He’s aware that some in the industry are wary of the technology or fear it will replace their roles. But he doesn’t think resistance will hold.
“AI is not going away, and resistance is not a strategy. My mandate is to encourage innovative thinking and to fundamentally reexamine how we work,” McKay said.
If McKay’s comments illustrate what’s possible in a digital‑first setting, Curt Hess (pictured left), US executive president at payments and infrastructure specialist Vitesse, highlights why many larger organizations struggle to translate AI pilots into scale.
When insurers, regardless of location, look to modernize, one of the first challenges is understanding their existing systems environment, Hess said.
Many organizations operate with a mix of legacy platforms, homegrown solutions, and heavily customized systems that have been built over time. If leadership teams don’t have a clear picture of how many systems are in place, how they connect, and who actually understands them, modernization efforts tend to start on shaky ground, he added.
That complexity is not just an internal issue. Insurance is a network business.
“Insurance operates as a connected ecosystem of carriers, brokers, MGAs, TPAs, reinsurers, and banks, all with different systems, incentives, and levels of technical maturity that still have to work together,” Hess noted.
An insurer can modernize its internal platform, but the moment a process extends beyond the enterprise, into delegated authority, shared risk, or third-party payments, progress is often constrained by the weakest or least-prepared link in that network, he added.
APIs have become the go‑to answer for integration, but Hess cautions against assuming that connectivity alone solves deeper structural problems.
“While APIs have made integration easier than it was in the past, they don’t automatically resolve underlying workflow, data, or process issues. New tools can connect systems, but they can’t compensate for foundations that were never designed to support real-time data flow or automation,” Hess says.
That’s the environment into which AI projects are being launched – and often stalled.
“This is where many insurers run into problems with ‘bolt-on’ approaches. Introducing AI or automation without addressing manual or disconnected processes underneath may deliver isolated improvements, but it rarely produces durable change. In some cases, it simply accelerates existing inefficiencies and creates new operational risk,” he warned.
Hess said that the most meaningful modernization efforts start with the core systems that support day-to-day operations, rather than with high-profile experiments. He noted that focusing on operational systems – like claims, payments, and financial management – often produces better results, as these areas are frequently overlooked but crucial for speed, control, and customer outcomes.
“Insurers that make progress typically treat modernization as an operating model shift rather than a technology upgrade. They simplify their systems landscape, improve interoperability, and align processes across internal teams and external partners.”
