Noel Sequeira and Tom Chamberlain: How data democracy is reshaping the insurance industry for the better

Artificial intelligence’s (AI) capability to automate simple tasks and assist users through more complex ones is becoming ever more pronounced –and insurtechs, such as insurance pricing firm Hyperexponential, are increasingly betting on the technology to drive a new wave of product innovations.

The insurance sector is primed to provide model use cases for many AI tools – underpinned by vast sets of unstructured data, the industry often sees skilled professionals, such as underwriters, performing manual data processing tasks to enable their core functions, such as pricing and analysis.

Speaking exclusively to Insurance Times at Hyperexponential’s annual conference, Hx Live 2025, on 10 July 2025, the insurtech’s chief customer officer Tom Chamberlain explains that while the insurance industry has always collected large volumes of data, the pathway from data to insight has traditionally been slow and retrospective, relying on Excel models and after the fact reports.

But AI enabled platforms and tools are changing this narrative.

He continues: “Now, underwriters can sit there and get all the information they need. All of this data sits behind the [AI enabled] platform – you’ve got the stuff that allows you to calculate the premium and you’ve got all of those risks that you’re adding into your portfolio. Everything gets stored and that data traditionally is not easy to get at.

“With AI tools, particularly our underwriting agent, you’re going to be able to interrogate [the data]. For example, an underwriter can ask ‘why is this premium so much higher than the last risk I wrote’ and [AI] will uncover what those rating factors are. Or they can ask ‘show me five similar risks to this within my portfolio’ and it will give that accessibility to the underwriter.

“To get this reporting in the past, you would probably need to wait two or three months and then go [and] make some decisions based on very, very historic data.”

Rate of adoption

The appetite for AI usage in insurance is strongest where the technology can automate repetitive tasks, but adoption becomes more cautious when it comes to decision-making actions – however, as Hyperexponential’s chief product officer, Noel Sequeira, explains, it all depends on the specific use case.

He tells Insurance Times: “It comes down to the use case. For repetitive tasks, customers are happy to adopt [AI] really quickly because the downside risk is quite low and you have a human-in-the-loop that can inspect how you’re ingesting data into the platform, for instance.

“When it comes to decision-making and using models to help you with reasoning, obviously you want to be a lot more careful because there are hallucinations, or [the technology] might not truly have looked up the right data set for that decision – and that could have bigger consequences.

“I have to say I’m pleasantly surprised by the rate of [AI] adoption and the willingness that customers show. I thought that the [insurance] industry would be a little slower moving, [that it] would wait to see how AI plays out in other industries, but I think there’s somewhere between 95% and 99% adoption of generative AI across the industry.”

Generative AI has the ability to create new content, including text, images, audio and code, based on patterns and structures from existing data.

Chamberlain agrees with Sequeira, adding that two factors are crucial for enabling a sense of comfort with AI workflows – data security and human oversight.

He says: “Data security is absolutely critical – that underlies everything. Then also the decisions you’re making based on [AI] – how much of this is human-in-the-loop and how much is going to be automatic.

“Particularly for the types of business we’re writing, the human-in-the-loop element is going to be really critical. We’re not looking at this to replace a judgement [from a human], it’s going to enhance that judgement.

“Eliminating repetitive tasks, that’s fine. But when we’re talking about decision-making, it’s really important for that human element to be in there, that you’re superpowering that person as opposed to just letting AI get on with it.”

Sustained innovation

The future of AI, as Sequeira sees it, is not just improvements in the raw power of any one foundation model, but the advancements that can be made by engineering the models to operate in a massively parallel fashion.

He explains: “Think about what it looks like to have 1,000 agents working in parallel.

“In software development, what we’re seeing is if you have a task, you might use OpenAI Codex – or Anthropic has a similar product – where you can give the same task to 20 agents and see which one comes back with the best answer. As you collapse the cost of inference, the cost of generation becomes lower, so you can have more concurrent attempts and find the winning one.

“That’s really going to have a massive productivity boost in multiple areas and that’s something that we’re really excited about. We’re at the frontier playing with what it looks like as it comes into focus.”