A R T I C L E
Cutting through the AI hype: delivering practical benefits for insurers from innovation
The next industrial revolution won’t be powered by steam or electricity - but by prompts, data, and decision intelligence. As generative AI moves from novelty to necessity, insurers find themselves at the tipping point of a transformation that’s less about replacing humans and more about reimagining how they work. This isn’t just a technology shift; it’s a redefinition of how value is created, risks are managed, and customer relationships are built. And the window for experimentation is narrowing.
According to Matt Hutchins, Partner and Global Insurance Lead at Capco, the differentiator will be speed to capability. ‘The people who will be successful are those who work out how to use these tools effectively,’ he says. ‘The sooner you start that learning curve, the better.’
While other industries sprinted ahead, insurance – typically cautious by design – has been quietly building its AI foundations. The industry’s AI frontrunners are finding that groundwork is paying off. From customer service to climate modelling, generative AI use cases are moving beyond proof of concept and landing in the field, fast.
What’s more, classical machine learning methodologies, such as natural language processing and advanced analytics, which have been deployed effectively in fraud, pricing, and risk modelling for many years, are now enjoying increased attention as a result of the huge spike in interest around generative AI. ‘That’s been a parallel side benefit,’ says Peter Wanda, Chief Data & Analytics Officer at Zurich Insurance.
It’s still the start of the journey for generative AI, of course, with companies exploring a range of possible ways to deploy the technology, with a focus on claims processing, customer engagement, fraud protection, and vulnerable customer identification, amongst other areas.
"It’s still early days, everybody is exploring and experimenting, but we are seeing some scaling now"
At AXA Health UK, the technology is already making an impact on the customer frontline, with contact centre agents using the summarization tools and recommendations to handle calls more efficiently, consistently, and personably. The end result is customers get a better call experience and faster access to the healthcare they need, says their Chief Operating Officer Philippe Duban.
AXA is already taking the lessons from early use cases and applying them to other areas. ‘The first use case is always the most difficult, because you have to set up the infrastructure, understand the data and learn the importance of a good quality prompt,’ says Duban. ‘But then you can roll it out, sweat the asset and develop more use cases faster, optimizing prompts very quickly in just a couple of weeks.’
Data: clouds, lakes and AI
Building these successful solutions requires some investment to overcome barriers such as data quality, business silos, and legacy systems, although these are not new nor specific to AI.
‘Legacy systems are a big obstacle but not insurmountable’ says Peter Wanda of Zurich, who advocates for a migration to a cloud-based architecture so companies can leverage cloud-based AI solutions. ‘Then you do not need to worry so much about the legacy systems.’
Indeed, much of the legwork required to modernize systems and improve data can be done by generative AI itself. The technology loves unstructured data, for example, making once impenetrable call notes or PDFs freshly useful so that new insights can be surfaced and workflows optimized.
Not all data is good data when it comes to training LLMs. This is where investments made years earlier to clean up data and store it in cloud-based data lakes can pay dividends.
‘We have a treasure trove of data going back sixty or seventy years, stored in our electronic archive for easy access and use. This data is managed 100% in line with our regulatory and customer data commitments, making it invaluable for simulations and risk analysis,” says Peter Wanda. He adds that the key is to focus not on the total universe of Data, but on the data required for your specific business cases. ‘Fixing all of the data is an unfixable task and nobody will fund it, so you need to enhance the data that is relevant to what you want to achieve. It’s an easy way to get ready for AI.’
Legacy systems: do it yourself?
When it comes to modernizing legacy tech, then generative AI can tackle coding that may have been developed 20 to 30 years ago, with the original skills lost to time, and help mine out the logic. In this way, companies can be empowered to rewrite and refactor code themselves, modernizing those core systems that are to be retained while other functions are replatformed in the cloud. It’s faster and cheaper, and should accelerate digital transformation across the industry.
‘Moving to the cloud and using AI can take 40% out of the work effort,’ explains Matt Hutchins at Capco, who advocates a mix-and-match approach for companies as they seek to reduce the costs and risks of migrating from their under-strain legacy mainframes.
‘For a complex estate, refactor some, rewrite some and replatform some,’ he says. ‘It’s a derisked, phased approach and we see a lot of CIOs exploring this now.’
Technical skills, for all
These models don’t work in a digital vacuum and still require humans to develop and train the models, monitor for bias and drift, and ensure use cases and outcomes are aligned with business objectives. Many of these skills and competencies are evolving as quickly as the technology itself, so companies are engaged in a war for talent with other industries seen as more cutting-edge than insurance.
‘It’s very difficult to attract the top data scientists because of course they want to work for Google or AWS and build the next LLM, but we don’t need that in our businesses,’ points out Peter Wanda at Zurich. ‘We need good data scientists who are keen to work in the business, and get excited by finding solutions and adding concrete value through their skills.’
Matt Hutchins at Capco echoes this, pointing out that insurance companies need to question whether they really want to build a workforce that looks like a technology firm.
‘This is a real consideration for leaders,’ he says. ‘What’s the ambition and where does the capability sit?’
According to Hutchins, for some this may mean a mix of hiring from outside, bringing in vendor expertise, and then seeking to upskill existing teams.
That appears to be the case at Zurich, where Peter Wanda reports that the existing data analysts in underwriting and claims are ‘being influenced by their environment and developing a real affinity to data’. ‘It is these people who will become your data scientists too,’ he says.
Like many insurers, AXA has established a data and AI academy to upskill its people and has engaged partners such as Capco to deliver interactive ‘promptathon’ sessions to support this initiative. Philippe Duban reports that around 30% of AXA UK’s workforce has interacted with the academy since its launch in 2024. ‘We want everyone to have a decent understanding of data and AI and its benefits,’ he says.
This will be key if the technology is to deliver on its promise, because successful scaling is not just about technology; it’s about people.
‘We have to make sure our people trust it, so they use it at scale,’ says Peter Wanda at Zurich. ‘If the same people that should be using the new tools are also the ones concerned about the future of their roles, then there’s a conflict of interest that needs to be addressed. This change management piece needs to be managed very well.’
The good news is that the right tools sell themselves by making a material improvement to job satisfaction. By undertaking many of the repetitive manual tasks, such as re-keying data or reading lengthy reports, workers are freed up to focus on higher value tasks, improving both their productivity and the customer experience. Indeed, frontline workers can be some of the biggest evangelists for the technology.
"When they have the right solutions, they love it,"
Back to front
When it comes to operationalizing AI, many firms feel more comfortable getting started in back office areas, away from direct customer exposure and where cost savings can be more obvious. It’s where firms can get the easy wins, gain the confidence of staff and ensure buy-in as the use cases move to customer-facing areas. Once frontline staff are equipped with AI tools, the impact can be transformative – and surprisingly quickly.
‘We have more than one third of our teams using generative AI every single day,’ says Philippe Duban at AXA. ‘We are already at scale.’
For customer-facing staff, call summarization has been one of the most powerful tools, delivering real impact for contact centre agents and customers.
‘Before our teams were making notes at the same time as being on the call, which interrupts the flow and their focus,’ says Duban. ‘Now they can give better CX and it also creates cost efficiencies. If you listen to a recording of a call from a year ago, and a call today, you would not believe the difference.’
Finding new opportunities
But AI isn’t just about enhancing what insurance companies already do. It’s about creating new opportunities, whether it’s new pay-as-you-live pricing tariffs or finding solutions that help build resilience in a changing world.
This can involve forging new relationships with customers. For instance, Zurich Resilience Solutions (ZRS) leverages AI and data-powered climate modeling and risk assessment solutions in their client engagements. This enables our Clients to understand their unique risks, identify adaptive measures, and build resilience into their businesses.
"Rather than the insurer just playing the risk transfer role, it’s about empowering the customer, reducing their risks, and closing the protection gap"
Humans, out of the loop?
Further ahead, there’s the potential for AI-powered bots and agents to act autonomously to complete tasks end-to-end. Gartner predicts that by 2029, agentic AI will autonomously resolve 80% of common customer service issues without human intervention, leading to a 30% reduction in operational costs.1
“AI agents are a significant step beyond the LLM-based chatbots and retrieval augmented chatbots already being deployed by insurers,” says Matt Hutchins at Capco. “Able to make decisions and take actions within set parameters, they can work with other tools and systems to help users achieve their goals with minimal guidance and oversight. There is a clear opportunity for insurers to integrate AI agents into each phase of the value chain, from initial underwriting and policy issuance to claims handling and ongoing customer service.”
Some start-ups and digital-only players already offer straight through claims processing, with customers sped from FNOL to settlement without touching a human. This is, however, still limited to simple cases, with more complex cover and claims likely to raise flags that need the attention of a human decision-maker. Here, AI can help, potentially spotting leakage that can compromise STP capabilities and ensuring more complex cases are quickly diverted to human experts, who have the tools to quickly review and intervene as necessary.
But while fully automated processes may be technically possible, insurers need to strike the right balance for the customer, both in terms of ensuring a fair outcome but also providing the empathy and compassion that may be needed in the moment. And, as Peter Wanda of Zurich points out, not all self-serve options are more convenient: uploading lots of photos from the scene of an accident can actually increase customer effort rather than reducing it.
Guard rails and audit checks
Human inputs are not just a nice-to-have, they are also mandated by regulators. Although there are automated tools to check for bias and model drift, the regulatory drive for explainability, transparency, and responsible AI means humans need to be firmly in the loop, checking that outputs are as expected and explainable.
For insurance companies, already heavily regulated, this oversight is part of being a responsible financial services firm.
“We definitely keep a human in the loop, we are not at the stage where we are fully comfortable with giving AI full autonomy. People are better at making more complex decisions and AI is a useful partner to them in this.”
Peter Wanda agrees. ‘Everything is moving so fast it is hard to see ten years ahead but certainly the next two to four years, we will see increased automation and humans will continue to be augmented by AI,’ he says. ‘But one thing I do not expect to change is the human in the loop.’
And one thing is clear: the winners of the cognitive industrial revolution future will those who embrace AI early. As Matt Hutchins at Capco noted earlier, the sooner companies start this learning curve, the better.
Those that delay risk falling behind in a landscape that’s evolving rapidly – not just in terms of tools, but in expectations from customers, regulators, and talent alike. Ready or not, the AI arms race is on.
Footnotes
Matt Hutchins,
Partner and Global Insurance Lead
Capco
Matt is a Partner and Head of Insurance, based out of Capco’s London office.
Matt joined Capco in January 2020 with a remit to grow the company’s insurance practice, with a particular focus on digital and data. Matt brings almost 30 years of experience in the consulting, insurance and retail banking sectors, and has worked with senior client leadership on a variety of strategy initiatives, transformation programmes and the delivery of partner-sourced managed services.
Matt joined Capco from HSBC, where he was Group Head of Transformation. Prior to that role, he worked at Accenture, where he was a Partner for 12 years, during which period he was also Head of Insurance for UK & Ireland for three years.
Matt holds a degree in Natural Sciences from Downing College, Cambridge University.
Philippe Duban,
Chief Operating Officer
AXA Health
Philippe Duban is Chief Operating Officer for AXA Health in the UK since January 2024, notably in charge of customer servicing, claims, Technology, Data and Operational excellence.
He had joined AXA in May 2021 as Transformation Director for AXA UK Retail.
Before AXA, he worked for HSBC in a variety of roles in Transformation, Strategy, and Business Performance/Finance, for both the Retail and Commercial banks. He had most recently been Global Head of Business Transformation for HSBC Wealth and Personal Banking with a focus on digitalisation and Agile.
He had previously worked for a French bank in Hong Kong, and at the strategy consulting firm Bain & Co, where he notably supported Insurance and Retail banks.
He is graduated from the INSEAD MBA and the French Business School ESSEC.
Peter Wanda,
Chief Data & Analytics Officer and Head of AI and Shared Services
Zurich
Peter is an innovative executive with 20 years of experience in leading and transforming organizations. He excels in driving business growth and transformation through technology, data, analytics, and AI strategies. Peter has a proven ability to deliver outstanding results in complex environments, holding executive roles such as CIO, CDAO, CTO, and COO across Insurance, Consumer, and Manufacturing industries.
A champion of ethical AI, distributed data architecture, data democratization, and advanced analytics, Peter enhances speed to market, customer experience, and profitability through innovative business models, efficiencies, and effectiveness. He strategizes and partners across all Lines of Business to modernize and transform Underwriting, Distribution, and Claims processes through data, analytics, intelligent automation (e.g., AI), and employee augmentation.
Peter strives for rapid reapplication across LoBs by maximizing platform approaches and accelerating adoption through cross-functional team models. Since 2021 Peter is Chief Data & Analytics Officer and Head of AI and Shared Services Zurich Switzerland.
About Insurance Innovators
Insurance Innovators is your one-stop shop for all the latest
insurance and financial services content from across the world. With a huge portfolio of conferences, webinars, executive roundtables and reports to choose from, Insurance Innovators provides our community with relevant and future-facing information.
To learn more visit insurance-innovators.com.
About Capco
Capco, a Wipro company, is a global technology and management consultancy specializing in driving digital transformation in the financial services industry. With a growing client portfolio comprising of over 100 global organizations, Capco operates at the intersection of business and technology by combining innovative thinking with unrivalled industry knowledge to fast-track digital initiatives for banking and payments, capital markets, wealth and asset management, insurance, and the energy sector. Capco’s cutting-edge ingenuity is brought to life through its award-winning Be Yourself At Work culture and diverse talent. To learn more, visit www.capco.com or follow us on Facebook, YouTube, LinkedIn and Instagram.
Stay current with Insurance Innovators
Subscribe to our newsletter to receive news, insights and special offers.