The decisions leaders make in the next five years will shape competitive advantage for the next thirty. That is not hyperbole. It reflects the accelerating, compounding impact of AI, and the reality that most organisations are still struggling to respond with clear focus.
At our recent EY AI in Insurance event, one theme ran through every conversation: the gap between what insurers say about artificial intelligence and what they are actually doing with it. EY research shows that 78% of insurers plan to have fully integrated agentic AI within two years. Yet only 7% have achieved meaningful scale. The distance between ambition and execution is where advantage is won or lost.
This is not another technology wave
What makes AI different is not the pace of change but the depth of it. We are moving from AI as a tool that assists people to AI as a system that can plan, reason, and execute end to end processes autonomously. The shift from Generative AI to Agentic AI is not incremental. It fundamentally changes how decisions are made across underwriting, claims, servicing, and distribution.
We are also moving from ‘bolt-on’ to ‘built-in’. The most progressive insurers are not adding AI to existing processes. They are redesigning operations around it. Cigna is deploying AI assistants that guide customers through claims and care journeys. Travelers is using AI to reshape cyber underwriting. Munich Re has embedded AI into its risk partnership ecosystem. These are not proofs of concept – they are early signals of a fundamentally different operating model.
A structural divide is opening
For years, commentators have talked about a performance gap between AI leaders and laggards. That framing undersells what is happening. This is no longer a gap – it is a structural divide, and it is widening.
Leaders are concentrating investment on a small number of high-value areas and driving them deep rather than spreading thin. They are learning faster, scaling what works, and compounding the advantage with each iteration. For those watching from the sidelines, that compound effect is the danger. The organisations pulling ahead now may not be caught.
AI will expand the industry, not shrink it
There is a persistent and misguided assumption that AI is primarily an efficiency story: fewer people processing more claims, lower operating costs, and a smaller industry overall. The economics do not support this view.
Economists describe a well-established pattern known as the Jevons effect: when the cost of something falls, demand increases rather than decreases. We are already seeing the potential for this in insurance. More risks are being priced, more scenarios modelled, more policies written. AI is expanding the volume and scope of work, not reducing it. Which is precisely why organisations that treat this as a cost-reduction exercise alone are missing the larger opportunity in customer growth, personalisation, and distribution expansion.
The question of empathy
Insurance is, at its core, a human business. People turn to insurers at some of their most vulnerable moments. The commercial logic of AI adoption means very little if the human experience of a claim or a service interaction becomes colder, more transactional, or harder to navigate.
This is the dimension that does not feature prominently enough in most AI strategies Automation can improve speed and accuracy and remove friction from routine processes. But speed without sensitivity is not a better customer experience, and it is not yet proven to drive better outcomes.
The most effective deployments use AI to free people to do what technology cannot: exercise judgment, read context, and respond with genuine care. In this model, AI handles the transactional and repetitive, while people remain the carriers of empathy. The goal is not AI that replaces human connection but instead to create more space for it.
That matters commercially too. EY data shows that 55% of customers worry that insurers will fail to comply with their own AI policies. Trust, once lost, is difficult to rebuild. The insurers who will win long-term are those who deploy AI in ways that earn confidence rather than erode it.
Where the real transformation is happening
The areas delivering measurable results today are claims, where AI is proving its value in automation, triage, and fraud detection, and underwriting, where improved risk selection is affecting both efficiency and loss ratios. But one of the most significant shifts we will likely see is in the front office.
AI is moving from back-office cost reduction to customer-facing growth. Personalisation, relevance, and experience are becoming the competitive terrain. That is where the economics of insurance change most profoundly.
What is holding firms back
Despite clear evidence of what is possible, 46% of AI initiatives stall between proof of concept and deployment. The causes are consistent: lack of focus, with investment spread too thin; data and legacy constraints, with 72% of insurers still dealing with siloed and poor-quality data; and organisational misalignment, with only 23% structured around AI-driven operating models.
The constraint is no longer the technology. It is the enterprise itself: the data infrastructure, the operating model, and the leadership choices about where to start and how deep to go.
What real transformation looks like
Real transformation is not a portfolio of AI pilots. It is not pilots installed across every function. It is end-to-end workflows redesigned around AI, decisions made faster and more accurately, new business models emerging, and a workforce that has genuinely changed in structure and capability.
The technology is proven. The leaders are emerging. The value is demonstrable. What remains is a leadership question: who is willing to reshape their business fundamentally enough to capture it, and who will do so while preserving the human qualities that make insurance worth having?
