Deborah Ritchie speaks to Hiscox UK’s Catherine Frost about how the insurer is approaching AI adoption and operational change, and how combining technology investment with continuous improvement principles is delivering benefits for both customers and employees


Hiscox Group’s latest results point to strong momentum from technology deployment and operational change programmes. What differentiates transformation that genuinely improves customer experience from transformation that simply cuts costs?

We’re in a period of real growth, and that includes our investment in technology. A core part of our strategy is based on our visibility of the customer through the policy. Whenever we think about investment, of course we have an eye on cost and on generating growth, but we’re also constantly asking how we make it easier for brokers and customers to trade with us. We use feedback from both to inform everything that we do in that respect.

Hiscox has a claims NPS score in the 80s, which is extremely high, and we use tooling and technology to improve and enhance the way we operate, but it doesn’t remove that human touch. In claims, we’re doubling down on fraud and recoveries technology, and making sure that wherever we partner – whether through our legal panels or other parts of the supply chain – that they share those same values.

On the placement side, we’ve recently brought in a case management system and a new contact centre platform, with the goal of making that entry into Hiscox as seamless as possible. Technology is helping us direct queries at the first point of contact through to the right decision-maker – removing that friction and effort from a broker or customer, while introducing speed, ease and faster turnaround.

Which investments have delivered the strongest operational return
on investment?

We’re really effective when it comes to understanding costs, but when it comes to understanding other associated benefits, we’ve been on a learning curve. We use the Microsoft estate for financial and MI reporting to create reports that run in the background without manual intervention. We have a really strong MI performance team who are brilliant at translating those metrics into something that makes sense for the business.

The greatest ROI has come more from process than technology per se. The best example of this has been a recent programme across our high net worth book – built on specific management principles, such as daily huddles, performance stats benchmarking against peer groups, and simple disciplines focused on problem-solving. When you’ve got the right continuous improvement team, the correct change management support, plus some really good leaders who take on those principles and run with them, embedding that process delivers brilliant results – both for customers and also for employees, because they better understand the business they’re working in and they know the parameters.

For Hiscox, 2026 is the year we are investing – in case management, our contact centre platform, and a rules and rating engine that will allow us to do more portfolio and algorithmic underwriting, which allows us to look at books of business rather than purely on a case-by-case basis, giving us a more consistent approach to pricing, and making us far more agile and responsive to whatever is happening in the market. That will of course bring with it costs this year, but it’s the right foundation going into 2027 and 2028, when it will form the bedrock for new pricing strategies, new products or improvements to our digital offering.

Do you see opportunities for acquisition-led growth?

Where we see an opportunity, and it’s the right thing to do, absolutely. We’ve recently announced our expansion into Italy through the acquisition of Lokky, a digital platform based in Milan. This move marks our first presence in Italy and aligns with our strategy to support SMEs through cutting-edge, digital-first products. We’ve just set up a branch and appointed a new CEO, who will be starting there soon.

How are you operationalising regulatory and risk management requirements without creating friction?

We’ve invested in a capability that helps us translate the different rules coming into force, and work out how to apply them. The regulators have a real focus on third parties – how we govern them, how we make sure they have the appropriate business continuity plans, and then make sure we have the right oversight. The teams we’ve built have helped us, over the past two or three years, to embed a realistic and not overburdensome environment where we have real confidence when we speak to the FCA or PRA, and where our approach is proportionate to our size.

On third parties specifically, we used to gather information on Excel spreadsheets, but have since partnered with an external supplier who has helped us build better processes around new and existing contracts. We’re now embedding some of these processes into our BAU. As we’ve grown our schemes businesses, for example, making sure that where we do provide delegated authority is now addressed upfront as part of the contracting process.

We’ve got a really nice operating rhythm around it now, where there are no surprises, but it doesn’t feel overly burdensome.

When deciding how far to roll out AI, how do you balance efficiency, the human touch customers need, and holding onto the talent that drives your success?

In my view, this comes down to how we use AI to enhance what our people are doing, to make their lives easier – so that in turn they can make our customers’ lives easier.

Over the last six months, for instance, we’ve given Copilot licences to everyone – which is quite unusual – and we see that as a real strength.

There are some great uses for AI across the insurance industry, such as in high-volume, smaller personal lines claims, for example a broken appliance: customers want to submit a photo through an app and receive an instant message confirming cover and payment. In those cases, you don’t need to speak to anyone. However, at the more complex end – such as a high net worth cyber incident – customers want a higher degree of certainty and a human at the end of the phone.

Placement is another great example. We have hundreds of thousands of enquiries coming through to us each year, and we’re now using AI upfront to ingest broker submissions, triage them, pass the clear yeses directly to the right underwriting team, and route the nos back to the broker automatically. That’s perhaps a less interesting part of the journey for an underwriter; what they’re skilled at – and trained to do – is look at a risk, understand its different facets, identify where they could offer additional cover, and price it properly. That’s where they add value.

Our group CEO Aki Hussain likens the advent of AI to Excel: when Excel arrived, everyone panicked and said ‘this is me and my job gone’. All it really did was change the way in which we operate. Now we wouldn’t think of life without it.

There are different layers to it. There’s the Generative AI layer – using Copilot, as we are, to summarise meetings and draft notes, creating new daily habits that make that layer of additional productivity more familiar. Then there are the bigger and more powerful purpose-built agents. One of our underwriters has been working with our data science team to build a tool that takes a broker submission, maps it against the underwriting licence and appetite, and comes back instantly with a recommendation on whether she can proceed.

Imagine being on the phone to a broker and instantly being able to say ‘yes, that’s in appetite and I can do that for you now’. That’s brilliant. In comparison to two or three years ago, when one of the frustrations for brokers was being passed from pillar to post, that sort of thing makes a real difference.

How do you expect AI capabilities to evolve across the industry?

I foresee dedicated squads sitting between the technical teams, which build the agents, and the business team, which can translate what we’re trying to achieve commercially into something buildable. That approach will fuse both the technical and commercial worlds.

Looking ahead, I can’t see a world where insurance will ever want to move away from that human touch – not just in claims, but in placement, where customers want to know they’re getting the best guidance for their unique situation. I would also expect customers will be much better informed before they contact us, so the conversation is more targeted and purposeful from the outset.

On hallucinations and other risks, I think the industry is showing the right amount of caution. Around 80 per cent of AI pilots currently fail, which isn’t a bad thing because the testing is useful. But I’m sure we’ll reach that inflection point where 80 per cent succeed. The insurance industry may not lead the way in AI, but in two to three years’ time, I think we’ll be surprised by how far we’ve come.

---