TechTalk: Walking the tightrope – the risks and rewards of tackling legacy tech

The UK general insurance sector has long wrestled with the burden of legacy software systems – ageing platforms and outdated infrastructure that prop up daily operations but restrict innovation.

While startups and greenfield players have the advantage of building on cloud-native foundations, many incumbents remain tied to decades-old estates, balancing the risks of disruption against the urgency to modernise.

At Lloyd’s, chief executive Patrick Tierney described the re-platforming of the market through Blueprint Two as ”mission critical”, but delivery has proved lengthy.

In the market’s half-year results 2025, Lloyd’s admitted that the project – intended to replace 100 mainframe applications with a cloud-based system – would not be completed before 2028, having been initiated as Blueprint One in September 2019.

Velonetic is the technology provider for the Blueprint Two project. Bob James, the firm’s chief executive, said: “What we are in the process of doing is modernising those 100 mainframe applications to 12 cloud-based applications.

“There’s a big risk to try to move from heritage estate to a brand new system,” he explained.

Calum Gibson, chief operating officer at Velonetic, also highlighted that the complexity of decades-long business plays a role in the difficulty of both initialising and completing these sorts of projects.

“We have claims, for example, that are going back multiple, multiple years,” Gibson explained.

”Somebody was talking to me about a claim the other day that went back to 1907. When you think about the complexity that the insurance [written at Lloyd’s] has, plus the length of tenure that some of these policies and claims have, that has an exponential impact on what it is we’re trying to do.”

Blueprint Two may be the most talked about digitalisation project in the UK insurance market, but it is emblematic of work that many insurers are undertaking. And, for many carriers, hesitation in committing to modernisation of their technology systems is both cultural and technical.

Rob Schumacher, co-founder at insurance agent Feather, noted: “Another reason why people are so stuck here is that a lot of the time, important people in insurance companies have, let’s say, five years left in a company – and they’re not going to start a risky project that they know might take 10 years. Change is scary.”

Natasha Bond, chief customer officer at Send, agreed with this assessment: “Historically, insurance has been a conservative, risk-averse industry. Maybe there’s a bit of an ‘if it ain’t broke, don’t fix it’ mentality. But I do see that shifting”.

She added that many executives recalled failed transformations: “We come up against fears of implementation risk. Insurance executives have seen or heard horror stories of failed system implementations costing hundreds of millions but you always hear about the bad stuff, never the good.”

This reluctance often translated into patchwork fixes rather than fundamental reform. Will McAllister, senior vice-president and managing director of EMEA at Guidewire, warned: “The biggest challenge we see in modernisation is the temptation to twist new technology to fit old process. So they really just want the new system to do what the old system did, but on more modern technology.”

Incremental fixes?

Despite delays and difficulties, most agreed that legacy modernisation was unavoidable. KPMG research, published in June 2025, showed that 35% of insurance leaders believed updating legacy technology architecture would deliver the greatest profitability gains over the next two years, ahead of process automation or other priorities.

The debate over whether to modernise incrementally or overhaul wholesale remained unsettled.

Bond said: “Modern tech, by hook or by crook, with some investment, can interface with legacy tech that’s on very old APIs and incrementally build towards modernising. Customers now want to see incremental improvement, rather than seeing anything as this kind of monolithic transformational change.”

James agreed, suggesting that wrapping and containerising legacy estates could buy insurers time. “There are insurance companies that are operating on 30–40-year-old heritage estates. What they’ve tended to do is hollow out some of the functionality that sits in these legacy systems with more modern applications. That’s probably a more prudent approach than just trying to change out everything it becomes quite risky and expensive.”

But McAllister argued that the integration problem could not be ignored. “You just can’t solve some of the data and integration challenges unless you take care of the core. Fundamentally, if you think about the innovation that data analytics and AI potentially can deliver to this industry, the industry won’t be able to take full advantage of that unless they solve the underlying core technology transformation challenge.”

McAllister added: “For those who potentially are still using 20–25-year-old legacy technology, that’s a huge challenge to think about how you get there, and they’re more likely to want to try to change incrementally, just because that feels like it’s more consumable.”

Opportunities and obstacles

There’s pressure to act sharpened as artificial intelligence (AI) reshapes customer expectations and underwriting practices. Deloitte’s global insurance outlook, published in January 2025, found that insurers now prioritise digital literacy and AI skills when hiring.

Yet many legacy estates remained incompatible. A report by Sprout.ai, published in May 2025, noted that older systems often lacked APIs, meaning even simple data exchanges required costly bespoke integration.

Matthew Soren Madsen, managing director at Accenture, highlighted cyber as a case in point: “Cyber is a brilliant example where it’s quite difficult just to put some sticky plaster or lipstick on a pig, because cyber touches so many different aspects of insurance. Taking an old platform, 40 years old, and trying to retrofit a cyber problem you can’t really do it.”

He added that the industry faced a ticking clock: “Some of these legacy systems are still what we call on-prem, which means it’s sitting on a server in the basement in the City. The system is 40, 50 years old. Everyone who coded it has either retired or is no longer around. At some point, you face the risk of it shutting down, and there’s no one around who can do anything about it. We’re about five to eight years out, I would say, from when you have to replace it whether you want to or not.”

Building for the future

Meanwhile, challenger brands, unencumbered by legacy stacks move faster. McAllister observed: “You’re seeing new startups or people starting up greenfield lines of business using modern technology from the get-go, just able to deliver a far better user experience outcome than they ever could if they were to do it on their legacy technology.”

This gulf matters, not only for operational efficiency but also for meeting customer expectations. In a briefing published in April 2025, EY warned that consumers now demanded fast, ‘mobile convenience and access to information’ – expectations that insurance, tethered to decades-old platforms, struggles to meet.

Gibson argued that transformation should be seen as foundational. “If I was building a house, I’d say I was redoing the plumbing and wiring, and then I get into doing the fun of making the rooms look more modern, up to date. But I shouldn’t do it before I fix the foundations.”

The question is no longer whether insurers could modernise, but how and when – and whether cultural barriers could be dismantled before technological ones. As Bond put it: “As digital natives pivot into senior underwriting leadership roles, maybe they won’t have the same emotional attachment to legacy processes.”