AI stalls on trust, not technology - 2026 ITC takeaways

This year, InsureTech Connect (ITC) Vegas offered no shortage of intriguing sessions, but a consistent message cut across them: AI progress depends less on model capability than on trust, governance and execution. The strongest conversations explored why programs stall, what makes an AI agent dependable, how redesigned workflows unlock value, and why insurers are shifting from paying for losses to preventing them.

Why AI stalls: scale, governance and people

A keynote on AI in insurance asked why so many efforts stall. The answer was less about models and more about execution. Carriers struggle to scale beyond pilots, stay focused on outcomes and establish the controls needed for safe adoption. Talent, governance, security and explainability still lag behind AI ambition. One session offered a practical test for every AI initiative: Does it work? Is the output useful? Is it properly governed? Do people use it? Can it scale? The message was clear: quality, control and predictability matter more than speed alone.  

A session on commercial lines analytics landed on a similar point from the human side: adoption depends on people. Models earn trust when users can understand what drives them, see what they produce and work with them inside familiar workflows. Independent agents and brokers may be especially well positioned to benefit because AI can strengthen customer relationships rather than replace them. That distinction matters: AI can automate tasks, but it cannot yet read the non-verbal cues and context that experienced professionals bring to customer conversations. 

What a trustworthy agent looks like

Several sessions covered agentic AI and the controls required to use it responsibly. Speakers framed the discussion around trust and transparency. Beyond accuracy, dependable agents need clear boundaries, predictable processes and human judgment for sensitive decisions. The goal is not autonomy at any cost, but autonomy that an insurer can understand, monitor and control.

The main agentic AI session, "The Agentic Insurer: From Ambition to Action," described a progression from AI as a thought partner, to supporting decisions, to acting within dependable controls. The first value comes from compressing and redesigning workflows, not simply adding another technology layer. Carriers should work backward from the customer problem and business impact, then determine where data comes from, how sensitive each decision is and when human judgment is required. A platform approach spanning conversational tools, coding tools and APIs can preserve flexibility, while role-based education and leadership training help the organization adopt the new processes at scale.

Core modernization offered another test of disciplined agentic AI adoption. Business-critical platforms carry complex, tightly coupled dependencies, making big-bang replacements especially risky. Speakers advocated a step-by-step approach: decouple systems, modernize deliberately and keep delivering throughout the transition. An AI agent can accelerate that work when it operates within a predictable process built on context, discipline and validation.

From reactive to resilient

A session on how technology is rewriting the rules of risk argued that understanding customer risk should be the starting point. Carriers want solutions to business problems, not a platform pitch, and the technology exchanged between customers and carriers rests on trust. The industry is moving toward prevention. Property owners care about natural disaster resilience but need incentives to act. The challenge for carriers is to innovate without eroding their business model.

Speakers shared that the opportunity is not simply AI but risk advisory: helping customers move from predicting losses to preventing them. That shift requires carriers to organize around the customer, understand risk more deeply and use data, incentives and trust to encourage preventive action. Success begins with the customer's problem, supported by industry experience and a clear point of view.

Rock climber Alex Honnold brought the theme home in his keynote. He does not eliminate risk; he prepares, takes calculated risks, and weighs the consequences of acting against the consequences of not acting. New risks will always emerge. Risk shifts rather than disappear.

Worth watching: Autonomous Vehicles and Systems (AV/AS). Multiple speakers cited industry and consumer reports that AV/AS are expected to grow significantly by 2030. Coverage is shifting away from traditional carriers, but there is an opportunity for those who move quickly and learn to transfer and manage AV/AS risk effectively.

Conclusion

Across the 2026 ITC Vegas sessions, the message was consistent: the winners will pair ambition with governance, predictability and measurable business outcomes. They will redesign workflows rather than layer AI onto broken processes, educate people for new ways of working and choose partners for durability and trust. Most importantly, they will start with the customer's problem and use technology to make insurance decisions, services and risk prevention more effective.

How WWT can help

The sessions pointed to the same gap: many carriers can pilot AI but struggle to reach production. That is where World Wide Technology (WWT) helps. WWT pairs insurance industry experience with a track record of building production-grade, trustworthy agentic AI, so carriers can move past the stall without trading speed for control. with a track record of building production-grade, trustworthy agentic AI, so carriers can move past the stall without trading speed for control.

In practice, that means three things. First, validating agents before they touch production: WWT's AI Proving Ground gives teams a production-grade environment to test and stress-test agentic workflows against real workloads. Second, building trust in from day one, with governance, security and auditability designed into each agent rather than added later, guided by WWT's AI Readiness Model for Operational Resilience (ARMOR), a trusted and deployed vendor-agnostic AI security framework developed with NVIDIA. Third, planning for scale across workflows, data, infrastructure and adoption, using WWT's Agentic AI Maturity Model to show where an organization stands today and what is blocking the next level.

Whether the next step is a candid look at where an AI program stalled or a plan to take an agent from proof of concept to production, WWT can help carriers turn what was shared at ITC into a working roadmap.