Early in my career, I sat on the carrier side of a major policy administration modernization. We did everything right by the project playbook: requirements gathered, milestones hit, go-live celebrated. And then I watched the same underwriters, advisors and service reps go right back to the same workarounds they’d always used.
That moment has stayed with me because it’s not an anomaly. It’s a pattern I’ve seen repeat across insurance carriers of every size and line of business. We have a tendency in this industry to treat technology transformation as a process migration. Move the old workflow onto the new platform to maintain compliance, then declare victory. What we rarely stop to ask is whether the process itself is actually serving the people on both ends of it.
The moment is real. The opportunity is massive.
Across underwriting, claims and service operations, the AI tools exist today to do things that genuinely weren’t possible five years ago. Telematics and IoT are opening new models for pricing and loss prevention. At the same time, customer tolerance for friction, the kind we’ve quietly normalized for decades, is eroding fast.
And yet, in boardrooms and steering committees across the insurance industry, I see the same gravitational pull toward the familiar: automate the existing process, satisfy the compliance requirement and move on. It’s the path of least resistance, and it’s also how you spend significant capital without meaningfully changing anything.
The irony is that the case for doing this differently has never been stronger, financially or operationally.
Human-centered design isn’t the soft choice. It’s the strategic one.
There’s a persistent assumption in insurance that optimizing for human experience and optimizing for return on expense are competing priorities. I’d argue they’re the same priority, just measured at different time horizons.
When you design experiences that actually work for underwriters, such as unified risk views, automated exception-flagging and real-time producer visibility, you compress cycle times. Underwriting that takes a month starts taking half a day. That’s not merely a customer experience win; it’s a loss ratio input. When claims handling works the way a member expects, with photo FNOL (First Notice of Loss), real-time journey orchestration and no re-explanation of context at every touchpoint, you reduce leakage, improve reserves accuracy and increase retention.
The same logic holds for the advisor and agent channel. Producers with fragmented tools and manual workarounds write less business, place it elsewhere and erode the distribution relationships carriers have spent years building. Fixing the advisor experience directly affects how many policies the company sells.
When you follow the thread, improved experience creates a cascade: waste declines, errors decrease, employee satisfaction rises, operational stability strengthens. Return on expense becomes an outcome of designing well, not a constraint that designing well fights against.
The portfolio mindset: not every problem needs generative AI
This is where I want to be direct with insurance leaders, because I think there’s a real risk of whiplash right now. Generative AI is not the answer to every question you’re trying to ask. Neither is agentic AI, nor automation, nor predictive modeling alone. Each has its place.
The advantage comes from a portfolio mindset: matching the right capability to the right problem, with the human experience as the organizing principle rather than an afterthought. Routine automation handles volume. Predictive models sharpen risk decisions and intelligent assistants reduce friction at the moments that matter. Generative AI adds value where language, synthesis and personalization are the bottleneck.
Instead of aiming AI at everything or using it for nothing because you’re too paralyzed to act, ask the better question first: what experience are we trying to create, and for whom?
Where to start
Insurance is a promise: that when something goes wrong, someone will be there. Every part of the value chain, from underwriting and distribution to service and claims, exists in service of that promise. The technology we build and the experiences we deliver should reflect that.
Instead of starting with a platform decision or an AI use case inventory, start with the people in your ecosystem. Take a clear-eyed look at where your current experience is falling short of the promise you're making, for employees, advisors and policyholders alike. That's where transformation with staying power begins.
