By Guy Moas, Senior VP of R&D, Sapiens
At the same time, we can improve risk calculations and create an error-free, continuously improving system. However, insurers need to adjust operating models and practices to capitalize on the AI opportunity.
AI is no panacea and is the polar opposite of the ‘one and done’ plug-in fix. It requires fundamental changes to achieve its impressively seismic effects. Indeed, when operating models are static, AI doesn’t fix weaknesses, nor is it even neutral. Instead, it has the effect of amplifying them.
Flawed underwriting logic becomes more apparent, inconsistent decisioning scales up and across, and compliance is voided, without even explainability and justification to mitigate the failure of governance.
As for life insurance customers, operational inefficiencies present themselves in the form of reduced and/or restricted modern, flexible products, slow onboarding, and delayed payouts. And, as life claims often carry a significant emotional burden, inconsistent or impersonal experiences damage insurer-customer trust, leading to lapsed policies and opting out. So, the benefits of AI are great, but any failure to re-engineer thinking and processes also comes with risk.
Individual failures are aggregated at the industry level into the global protection gap, valued at over $900bn and a figure that represents not just a commercial shortfall, but a failure of the industry’s core social purpose.
The market opportunity
Stronger investment margins have improved insurance performance, but they require a reset. Long-duration liability management and portfolio positioning must adjust to a steeper yield curve, while renewed demand for savings and retirement products is lifting volumes. To absorb that growth, operating models need scalable governance, solid ALM, and tighter capital planning, all to capture the expected uplift through 2027.
The challenge for insurers is that their operating models are not dynamic enough to give them the agility they need to react in real time to such volatility. This creates what we call the Performance Gap, of which there are five main indicators – operational, financial, compliance, decision-making, and speed to market.
The outperforming companies in the coming years will build dynamic operating models capable of absorbing change, aligning decisioning, and sustaining performance across cycles.
The role of AI
McKinsey estimates that automation and AI can reduce claims-handling costs by 25 to 30%, and compress processing times by 40-50%. Forrester’s 2026 forecast adds that AI and automation will improve expense ratios at the top 50 insurers by two percentage points, but only for carriers that have scaled beyond pilots into full production.
So, how are insurers making use of AI? It’s clear to me that the world is moving fast on LLMs and, sooner or later, we will see more adoption of embedded models unique to insurers as price points fall and as models mean that use cases become more immediately relevant.
But it’s also important that organisations plan for softer factors. One area to consider here is the impact on skills. Insurance veterans are used to specific role-based skills such as claims handling or underwriting. But AI acts as an accelerator here, flattening roles and steamrollering discrete processes.
So, insurers need to think about how they encourage people to be all-rounders capable of orchestrating and adding value to end-to-end processes.
Time to market is also key to success in modern insurance and AI is an enabler of this. But it’s also an enabler of speeding up customer activities.
Underwriting, for example, is still extensively based on lengthy, time-consuming questionnaires and check-list questions. AI is a massive force for collation, number-crunching and interpretation. We still need the human in the loop, but we can get to decisions faster and delight customers in doing so. Similarly, on the claims side, we see massive potential in how to initiate claims and in compliance management because we can narrow down the level of potential misunderstanding or misalignment.
AI is only going to become more important, particularly now we can make use of so-called RAG and CAG tools to accelerate integration of data to and from LLMs, via query retrieval and pre-load caching techniques, respectively. That means, practically, that a junior underwriter, for example, can quickly cycle through harnessing key information, gaining managerial approval and completing the end-to-end process. Administration becomes a sign-off action and manual inputs are massively reduced.
We are reaching the point where we are mapping all the key data in the insurance organisation, but also, just as important, seeing what is missing so an external source may be plugged in or another action taken.
Insurers can’t be bystanders, and they can’t just rest on their laurels and admire their work in being benign early adopters of AI. The winners in the sector will be those who deploy at scale, and make the necessary adjustments to people and processes to take advantage of one of the great technology-enabled waves of change in our lifetimes.
|
