The next phase of AI and digital integration

Artificial intelligence (AI) has become a dominant topic at recent re/insurance gatherings, but few insurance sector companies are realising its full potential, and many are barely scratching the surface.

 For technology providers working at the intersection of underwriting, claims and data science, the focus now is less about experimentation and more about embedding AI into the practical mechanics of insurance.

“AI is going to be a tool that people use to improve the job that underwriters do,” said Dani Katz, co-founder and director at Optalitix. “It will make it easier to do certain things like quote ingestion, which is a very low-value task.”

Optalitix has upgraded its Quote platform to help insurers and reinsurers move from spreadsheet-based pricing models into a cloud-based environment, using APIs to manage operations. The firm’s model engine converts data held within excel spreadsheets into APIs, turning them into front-end workbench tools that underwriters can use.

“The elevator pitch,” Katz said, “is that we convert existing pricing models into a front-end system that captures all the data, manages the business and removes double capture.”

As underwriters prepare for softer market conditions, the emphasis is shifting from rate increases to precision and discipline.

Katz explained the newer features allow reinsurers to upload premium and claims bordereaux, validate them using AI, generate loss forecasts and model distributions for quota share or excess-of-loss covers within a single system.

“These features allow senior management to more easily manage and optimise their portfolio over time. When the rain starts to fall, you need to manage your risks far better than before,” he added.

For insurance technology service provider Xceedance, its approach to AI is increasingly connected to issues of costs, data governance and operational efficiency.

 “Everyone wants to go through digital transformation and embed AI into their processes,” said Isabelle Clausner, client executive for southern Europe, acknowledging the buzz around technology for this year’s re/insurance sector conference season.

“But the recurring pain point is data, being able to collect data across legacy systems and then structure it so that AI can actually add maximum value.”

Gavin Lillywhite, Xceedance’s operating leader for the Europe, Middle East and Africa, said this challenge has become more pressing for re/insurers in a softening market.

“AI can help augment underwriting so that companies can be more targeted and manage softer market conditions,” he said.

“If you don’t meet the newer demands of clients, whether that’s for frequency perils such as wildfire, or intangible risks like crypto currency, then all you do is end up competing for more of the same, and that drives the race down [in pricing competition],” he continued.

Clausner added that only around one in ten insurers currently claim to be embedding AI in workflows.

“That’s probably not the whole truth, it’s just that employees aren’t always being honest about how they’re using it,” she said, suggesting some individuals are losing patience with the pace of change at their employers.

Joining the dots

If AI is at the front end of transformation, integration is often focused on the harder-to-change infrastructure that lies behind it, and this devil in the detail has contributed to the holding back of generations of Lloyd’s and London market digital transformation initiatives.

Despite the complication and nuance, few issues attract stronger opinions than the need for common data standards and connectivity across the market’s disparate systems.

For James Willison, managing director at Web Connectivity Limited (WCL), interoperability is the missing link that can make digitalisation practical.

The business, part of US insurtech parent Zywave, has been deeply involved in projects that connect brokers and re/insurers through structured data, including long-standing participation in the technology standards focused Rüschlikon Initiative.

“Back in Bermuda we got about twenty companies together,” Willison recalled. “We said, ‘This clearly isn’t the way to go on the accounting side.’ The Rüschlikon Initiative had been very successful using common data standards. So why not something consistent on the placing side?”

That discussion has led to an interoperability hub allowing reinsurers to see all their submissions from multiple broker platforms in one place, aggregating data, documents and quotes regardless of the systems used upstream.

“It becomes irrelevant what technology the broker is using,” he said. “The reinsurer can export that data into their own workbench and start responding directly.”

This kind of standardisation is already transforming premium and claims reconciliation. Through Rüschlikon’s data protocols, brokers and carriers can now share ledgers in real time, pre-authorising cash flows and cutting reconciliation times from weeks to hours.

“Reconciliation rates suddenly went up to essentially 100% because you’re pre-authorising what money needs to flow,” Willison said.

The name, he noted, comes from Swiss Re’s innovation centre on the shores of Lake Zurich, where brokers and reinsurers first met to solve “a cottage industry of backwards and forwards spreadsheets”.

Today, the same principles are being applied to placing and accounting data, aiming to build a wider digital ecosystem for re/insurance.

Optalitix’s Katz also sees integration as the bridge between innovation and adoption.

“Our clients pick the partners they want us to work with,” he said. “We provide the plug-ins. It’s all about enabling them to use the AI ingestion tools and data systems they prefer – safely and efficiently.”

Integration is equally central for Xceedance, Lillywhite emphasised. The company partners with core technology providers, such as Duck Creek and Guidewire, helping clients to implement, upgrade and migrate systems to the cloud.

“We want to move beyond traditional managed services,” said Lillywhite. “That includes implementation, upgrade management and managed IT services – effectively everything that helps insurers run efficiently while keeping focus on their core business.”

Digital transformation: progress and pitfalls

Even as AI and interoperability advance, the digitalisation state of the London market remains uneven. Lloyd’s Blueprint Two has drawn both optimism and frustration in equal measure, sources broadly agreed.

Willison said that although the “build” phase of the project has been declared complete, the testing required for go-live could make timelines “challenging at best”.

He cautioned that replacing the market’s decades-old mainframe systems with modern infrastructure has proved more complex than expected.

“You’ve got old infrastructure that still needs updating,” he said. “But they’re trying to replicate every single nuance of forty years of bespoke messaging. Many of the people who understood best how those systems worked have since retired.”

The project’s goal, to reproduce existing functionality on new architecture without disruption, risks being a paradox, he suggested.

“It makes it difficult, because the best you can hope for is that people don’t notice any change, and that’s an awful lot of budget and resource for something you don’t notice.”

Still, most technologists agreed that momentum behind transformation is real, even if the path gets muddled at times.

For Xceedance, the shift is now toward what Clausner called “an AI-led managed services model”, supporting clients not only in operations but in governance, regulation and risk.

“It’s about putting in place the right framework, data governance and AI governance, so they can roll it out safely for their organisation and their clients,” she said.

That includes navigating emerging regulation such as the European Union’s AI Act, which will likely define “best practice” for model transparency and bias prevention in the years ahead.

Bias may prove a bigger long-term threat than data breaches, observed Lillywhite.

“It’s easy to train a model to become more biased,” he said. “And if everyone’s using the same open model, that bias spreads to everyone’s use of AI. That’s the real contagion risk.”

Clausner stressed that insurers’ sensitivity over their own data sovereignty – fearing feeding their data to off-the-shelf AI vendors – is shaping how the technology gets deployed.

The ultimate risk of this going awry would be costly legal, regulatory and reputational consequences for any firm found culpable.

“We create sovereign AI platforms that allow clients to do what you can do with ChatGPT, but within a closed environment,” Clausner said.

“The aim is that sensitive data stays within their environment, and the small language models are trained for specific tasks,” she added.

Pragmatic evolution

Conversation has shifted from grand visions of transformation to practical, incremental improvement.

As Lillywhite put it: “We’re not trying to revolutionise. It’s more evolutionary, improving processes and making them better.”

For Katz, pragmatism is appropriate for a softening re/insurance market facing renewed competition.

“In a hard market you can afford a hazy idea of pricing,” he said. “But as it softens, you need to know where you’re willing to go to win business and at what point it starts to hurt you.”

Willison echoed that the technical plumbing of the market ought to match underwriters’ talk of discipline.

“Ultimately, digital transformation isn’t about replacing relationships. It’s about making sure the data behind those relationships flows seamlessly,” he said.

Clausner added: “Technology alone isn’t the answer. It’s about combining data, structure and training, safely and intelligently, to create real value.”