The potential of generative AI has caused a widespread race across the insurance sector. Firms are embroiled in various pilots and tests to see where the most value can be found, but where is the real value coming from?
Firms have spent the past few years experimenting with generative AI, but the time for pilots is coming to an end and firms are looking to move into real deployment of the capabilities. A recent report from Celent on the state of generative AI within insurance found that 48% of insurers it surveyed are currently in production with generative AI. It anticipates this to exceed half of firms by the end of the year.
As firms move ahead with their deployments, this raises the question of where these tools are creating value today.
Speaking to FinTech Global, Matthew Twist, vice president, EMEA at Earnix, explained, “Every technology wave arrives with the same promise: it’ll change everything. Insurance has learned, sometimes the hard way, that technology only matters if it improves decisions. The interesting conversations have changed. A year ago, everyone wanted to know what generative AI could do. Today, they are asking where it delivers measurable value.”
The answer, according to Twist, is by putting AI tools in front of experienced people to help them reach better decisions more quickly. Rather than building a new dashboard for underwriters, he noted, they are better served by a tool that can digest 200 pages of submission data and not miss anything important. Similarly, claims handlers don’t need to be completely replaced by AI; rather they need a tool that can collate relevant information so they can make a better decision. “That’s where I’m seeing genuine traction.”
Twist also warned of a risk firms often fall into. The heightened hype surrounding generative AI creates a fear of missing out environment. This results in firms rushing to find problems an AI tool could solve when there are much better, and cheaper, alternatives to fixing the problem.
He said, “I’ve seen organisations reach for generative AI simply because it’s the technology everyone wants to talk about. Sometimes a rules engine or straightforward workflow automation is exactly the right answer.”
Elsewhere, Melanie Hayes, co-founder of KYND, believes that some of the most tangible value is not from how an insurer uses AI, but how their clients use it. She said, “Some of the most tangible value we are seeing sits away from insurers’ own deployments, in understanding what AI adoption among insureds is doing to the risks already on their books. AI is now embedded across businesses of every size, often through third-party tools rather than any deliberate strategy, and much of it never appears on a proposal form.
“Carriers, brokers and MGAs writing AI cover tend to describe a similar picture: exposure building steadily, largely undeclared, and concentrated in a handful of widely used models. It looks like an accumulation question, and cyber risk analytics that map a live technology footprint can help bring it into view at the point of underwriting.”
When it comes to finding the insurance processes that stand to benefit the most from the implementation of AI, both Twist and Hayes agreed that its biggest value will be on areas that rely on data.
Hayes said, “Suitability probably has less to do with the process itself and more to do with the data feeding it. Generative AI is proving useful in high-volume, language-heavy work: triaging submissions, summarising long reports, drafting client communications.”
However, these are areas where source material already exists and models just help to make the data easier to work with. Data that is incomplete or outdated cannot benefit as much.
“The decisions that matter most in cyber underwriting, from risk selection through renewal to accumulation management, rest on a different foundation: whether the underlying data is current and complete. A model summarising a month-old scan will produce a fluent description of a risk that may no longer exist, and it will do so convincingly.
“The same applies to AI exposure itself. If it is difficult to see which insureds depend on which AI services, generative capability alone cannot fill that gap. In practice, the processes best suited to GenAI tend to be the ones already built on live, complete risk data, with everything else benefiting once the data foundations are in place.”
In the same vein, Twist explained that generative AI will be best utilised in the processes that require the assessment of established data.
“Insurance is full of information that’s difficult for humans to consume quickly. Broker submissions, engineering surveys, policy wordings, claims files, customer correspondence – that’s where generative AI is genuinely useful because it can make sense of unstructured information at a scale no individual can match.
“But it shouldn’t be making the decision on its own. Insurance decisions affect people’s livelihoods and businesses. There still needs to be accountability.”
As firms move beyond pilots and into deployment, there will be those that succeed and those that don’t. For Twist, the winners will be defined by helping clients make better decisions before a losses occur.
“The recent return to profitability across many insurance markets should not create complacency. Insurance is cyclical, and sustainable advantage rarely comes from rate increases alone.
“The brokers and MGAs that outperform will combine specialist expertise with technology to gain a clearer, more dynamic view of risk, pricing and exposure. They’ll use AI and data to augment experience, strengthen underwriting discipline and help clients understand emerging risks before they become claims. Those businesses will become trusted partners in resilience, rather than simply distributors of insurance products.”
Copyright © 2026 InsurTech Analyst
