On this episode of FNO: Insuretech, hosts Rob Beller and Lee Boyd sit down with Kyle Nakatsuji, CEO of Clearcover and founder of Dearborn Labs, for a wide‑ranging conversation about what it actually takes to deploy AI inside an insurance business and why the industry has quietly reached an inflection point.

After nearly a decade building production AI systems within a regulated insurance carrier, Kyle brings a grounded perspective to a moment that feels increasingly overwhelming for insurance leaders. The discussion spans lessons learned at Clearcover, the limits of traditional technology implementations, and why recent advances in AI have changed not just tools, but expectations around how insurance organizations can operate.

From building AI in house to sharing the playbook

Much of the conversation starts with Clearcover itself. Over ten years, the company invested heavily in building AI that worked across claims, customer service, underwriting support, and distribution. That work delivered measurable efficiency and cost improvements, but it also came with hard lessons about what it means to make AI work reliably inside a live insurance environment.

Kyle explains that for a long time, that experience only benefited one company. As conversations with other carriers, MGAs, and agencies increased, a pattern emerged. Many organizations were trying to adopt AI but were struggling to move beyond pilots, vendor demos, or strategy decks that never translated into business results.

That realization ultimately led to the creation of Dearborn Labs, not as a traditional software vendor, but as a way to help insurers figure out what is actually worth building, how to implement it, and how to measure whether it worked.

The AI overwhelm executives are feeling right now

A recurring theme throughout the episode is executive overwhelm. Kyle describes conversations with leaders whose boards are asking pointed questions about AI while their inboxes are flooded with vendor pitches, each promising transformative results.

The problem, as Kyle frames it, is not a lack of tools. It is the inability to determine what matters, what can wait, and how to move from experimentation to production in a way that delivers real outcomes. Rob and Lee echo this sentiment, noting that nearly every executive today is being pressed to articulate an AI strategy without clear guidance on how to separate signal from noise.

This tension sets the stage for a deeper discussion about why traditional approaches to implementation are breaking down.

Why traditional implementations no longer make sense

One of the most striking parts of the conversation centers on how implementation itself is changing. Kyle argues that the era of multi year, people heavy technology programs staffed by hundreds of consultants is no longer necessary or practical.

Instead, modern AI systems allow small forward deployed teams to orchestrate large amounts of work through agentic tools. One experienced engineer, using the right models and workflows, can accomplish what previously required entire departments. This shift does not remove humans from the equation but changes their role from execution to oversight, orchestration, and judgment.

Importantly, Kyle emphasizes that insurance is a complex adaptive system. Claims, underwriting, customer service, and finance are deeply interconnected. AI systems that operate in silos may optimize one function while unintentionally creating issues elsewhere. The future, he argues, lies in cross functional agentic systems that understand the business context, not just isolated tasks.

The week everything changed

The conversation also touches on a specific moment when AI progress stopped feeling incremental and started feeling exponential. Kyle points to a short window when Anthropic released major new Claude capabilities alongside a wave of agent based experimentation across the industry.

During that week, teams were not just improving productivity but rethinking how knowledge work itself could be done. Agents were being built to collaborate with each other, coordinate work, and even communicate autonomously. The message was clear. What felt like a gradual evolution suddenly became a step change.

Kyle describes this as the moment many leaders realized that being six months behind could very quickly become being six weeks behind, and that waiting was no longer a neutral option.

Moving from strategy to results

Throughout the episode, Rob and Lee repeatedly return to a core question. How do insurance organizations move from talking about AI to actually changing how the business operates?

Kyle’s answer is consistent. Technology alone is not enough. Real change requires understanding existing workflows, aligning incentives, supporting adoption, and measuring outcomes over time. From claims copilots to AI driven customer interactions, success depends on how tools are introduced and embedded into daily work, not just how powerful they are.

Examples from Clearcover highlight that adoption often improves when teams are involved early, trained through real use, and incentivized to experiment responsibly. AI succeeds when it becomes part of how work gets done rather than something bolted on from the outside.

A shift in how insurance leaders should think about AI

As the episode wraps, the discussion zooms out. Kyle reflects on how AI will continue to evolve alongside insurance operations. In the near term, the biggest risk is doing nothing. In the longer term, competitive advantage will come from organizations that learn how to create environments where AI agents can operate effectively under human oversight.

Rob and Lee close by noting that transformation in insurance rarely looks dramatic from the outside. The most impactful changes often happen quietly, inside workflows, decisions, and infrastructure. Episode 304 is ultimately a conversation about how that kind of change is beginning to accelerate and what it will demand from leaders willing to engage with it.