Artificial intelligence has moved beyond experimentation. APAC insurers are reporting tangible gains from applications across claims, pricing, underwriting and customer service.
Yet a tension persists.
The most successful deployments so far tend to share a common characteristic. They improve an existing process rather than fundamentally change how the business operates. But scaling those applications across functions, markets, and legacy systems is proving to be more challenging.
This is becoming even more significant as insurers turn their attention to agentic AI, which promises to coordinate tasks and decisions across entire workflows.
Increasingly, insurers are shifting their focus from looking at what AI they can build to asking the question: ‘How should we redesign the business so AI is at the core of how we operate?’
Xinfa Cai
Group Chief Innovation Officer of AIAMind the gap
The applications vary considerably between insurers, reflecting the different ways they are approaching the technology. At AIA, the focus has included using AI to support and develop its agency workforce.
“AI is transforming our Premier Agency [the company’s core agency distribution channel], freeing up time for what matters most: trusted, personalised advice,” says Xinfa Cai, Group Chief Innovation Officer of AIA. “We are using AI to scale high quality coaching and capability building, alongside intelligent recruitment and project prioritisation. Tools like ‘AI Role Play’ enable agents to practise real customer scenarios, receive immediate feedback, and continuously improve. For agency leaders, AI also enables stronger talent potential identification, deeper team insights, and more effective leadership.”
Generali offers another example of AI moving beyond individual use cases, with applications being deployed across multiple markets and business units.
“We are already seeing measurable value across multiple markets and business units,” says Laurent Crouet, Chief Transformation and Operations Officer for Generali Asia. “Examples include geospatial intelligence, AI pricing engines, and health claims automation.”
Generali claims to have improved its combined ratio by around two percentage points across each of the seven business units where AI pricing has been introduced. AI has also driven a 40% increase in straight-through processing for claims across six business units in healthcare.
Laurent Crouet
Chief Transformation and Operations Officer for Generali AsiaCrouet says that the most advanced use cases have emerged as the ones with high data availability, repeatable processes and clear business incentives to increase operational efficiency. These include supporting the claims process, strengthening pricing models and automating customer interactions.
Such comments reflect a broader sentiment across the industry. Many insurance leaders are no longer focused solely on proving that AI works or capturing isolated quick wins. They want to move towards more coordinated, enterprise-wide transformation that can deliver sustained impact at scale.
The experiences of firms such as AIA and Generali highlight a broader evolution in AI strategy. Early deployments have demonstrated that AI can deliver tangible benefits in areas such as claims, pricing, distribution and customer service. The next challenge is scaling those successes beyond individual teams and markets.
As insurers seek to make AI a foundational capability rather than a collection of disconnected initiatives, attention is increasingly shifting towards the organisational, data and technology foundations required to support enterprise-wide adoption.
“I think it is probably fair to say that the industry is further down the maturity curve than they would probably like to be,” says Vanessa Maher, APAC Chief Operating Officer for Liberty Global.
“At Liberty, we’re pivoting from bottom-up ideas and short-term wins to a more strategic approach where we define the key business problems we need to solve, and then use AI in a reusable way across multiple markets to improve ROI and speed of deployment.”
However, for all the hyperbole surrounding the new technology, some insurers admit that productivity gains have fallen short of expectations.
Beat Kramer Mölbert
Swiss Re’s Group AI Enablement LeadImplementation challenges
Early AI deployments have demonstrated what the technology can achieve. The next challenge is embedding those capabilities across complex insurance organisations at scale, where success depends on far more than the performance of the AI itself.
“The biggest constraints are not the models themselves, but the underlying data, architecture and operating environment,” says Beat Kramer Mölbert, Swiss Re’s Group AI Enablement Lead. “Insurance is still shaped by fragmented legacy systems, siloed data and highly manual workflows, making it hard to scale AI beyond pilots.”
In APAC, these challenges are often amplified by the diversity of insurers’ operating environments. Larger regional insurers may run multiple core platforms spanning different markets, product lines and acquired businesses, while regulatory requirements and underwriting practices vary significantly between jurisdictions. Scaling AI therefore depends on whether data, architecture and governance can support consistent decision-making across this complex landscape.
Vanessa Maher
APAC Chief Operating Officer for Liberty GlobalA strategic shift
The next stage of AI adoption is a shift in mindset. Rather than evaluating individual use cases in isolation, organisations are increasingly asking how AI can create value across functions, markets and business processes. This requires moving beyond a project-based approach to one that treats AI as a long-term strategic capability embedded within the business.
“We initially focused heavily on use cases but ultimately come to the view that this was not delivering the return on effort or the level of transformation we expected from AI. As a result, we took a step back and asked ourselves: what is the best way to approach this new way of working?” says Maher from Liberty.
This shift requires closer alignment between business, technology and operations teams, alongside clear governance and a broader view of how AI can create value across the enterprise.
“This has become less of an AI challenge and more about thinking in a totally different way,” says Matt Reilly, Chief Operating Officer for APAC at Zurich Insurance.
“We’re at the point now where we’re taking a broader view of AI, looking at where it can create value across the business and thinking about deployment in a much more holistic way. This doesn’t mean we’ve got huge programmes with massive investments that are going to take years. It’s about building a broader vision that transforms how we work.”
Reilly says it is natural for insurers to begin with targeted use cases to understand what AI can do. The challenge is recognising when those early successes need to evolve into a broader enterprise strategy.
Matt Reilly
Chief Operating Officer for APAC at Zurich InsuranceZurich’s commercial insurance business illustrates this approach.
Rather than launching a single, large-scale transformation programme, the company has defined a long-term vision centred on an agentic AI-driven operating model. Over time, this could involve more than 100 AI agents performing underwriting, servicing and claims-related tasks across the commercial insurance value chain. Instead of attempting to build everything at once, Zurich is deploying these capabilities incrementally by business line, task, and market.
This enables the company to deliver early benefits while steadily building towards a broader transformation of how commercial insurance is delivered.
Execution matters
Leadership commitment, effective governance and a clear vision are vital if insurers are to scale AI successfully.
“If delivered well, the return on investment for AI use cases can be pretty good, with a payback and healthy returns after two years,” says Bernhard Kotanko, a senior partner at McKinsey, who advises insurers on AI transformation. “The problem is that many firms have been in a rush to prove certain use cases. It is quite frankly irrelevant how many use cases a company has. These are just creating costs.”
For Kotanko, two factors greatly enhance the chances of success. The first is strong buy-in from the very top of the company. The second is an effective “workflow orchestration” layer that breaks complex tasks into manageable steps and coordinates decision-making across them.
Without the right organisational foundations, insurers risk creating isolated AI initiatives that fail to realise their full potential.
Bernhard Kotanko
Senior partner at McKinseyAI as a redesign trigger
The emergence of agentic AI is bringing these organisational challenges into sharper focus. Rather than automating individual tasks, agentic AI enables multiple models and agents to work together across end-to-end workflows, reshaping how decisions are made and executed. This shifts the focus from deploying AI in individual functions to redesigning entire business processes.
“The real opportunity is not a single model performing a single task, but multiple capabilities working together across the value chain,” says Chirag Shah, CEO Asia & Insurance at SAP Fioneer.
But as AI moves from task-level assistance to workflow orchestration, the quality of the underlying foundations becomes more important.
“If the underlying data, architecture and process design remain fragmented, agentic AI can add further integration complexity rather than reduce it,” says Shah.
The insurance sector also places clear boundaries on where AI autonomy is appropriate. While agentic systems can support more complex decision-making and process coordination, responsibility for core business logic, control and accountability must remain anchored within the organisation, says Shah. In areas such as underwriting, he believes AI is more likely to augment human judgement than replace it in the near term.
“Maturity-wise, the industry is still at an early stage. Interest is high, and some insurers are beginning to define broader agentic operating models, but the move from isolated use cases to coherent, end-to-end orchestration across core insurance processes is only just beginning,” says Shah.
Chirag Shah
CEO Asia & Insurance at SAP FioneerFrom use cases to business value
Insurers are starting to shift from asking the question, ‘what AI can we build?’, to asking, ‘how do we improve loss ratios, speed and cost across markets?’.
“We’re likely to see a much more horizontal journey, organised around business domains, where experts from different disciplines work together,” predicts Kotanko. “In practice, this means a tighter integration between business and technology ownership – what some refer to as ‘two in the box’ – and ultimately a model where they operate as a single, unified team.”
Mastering the technology has only been one half of the battle. The next stage is moving from proof of concept to an operating model where AI can deliver sustained business impact.
“AI only creates value when it is embedded into how decisions are actually made, with humans firmly in control, clear accountability and cross-functional collaboration between business, operations and technology,” concludes Swiss Re’s Mölbert.
“What we have seen time and again is that the real challenges lie in data readiness, organisational trust, and the ability to integrate AI meaningfully into existing core systems and workflows,” says Shah. “Insurers will only move beyond isolated pilots when technology, business ownership and operating design come together.”
