This piece is by Christian Preece, Insurance Director at Evident
In recent years, insurers have made significant progress in embedding AI across their operations.
Investment in AI has accelerated over time, with specialist hiring growing by a third across the past year, even as overall headcount remains broadly steady. Meanwhile, to accelerate the development of frontier capabilities, over the past three years the 30 insurers in the Evident AI Index for Insurance have completed more than 326 investments in technology and data companies, with AI accounting for almost half.
Rather than AI being confined to innovation teams and trials, many of the sector’s largest firms are now deploying AI across underwriting, claims, customer service and internal operations. So far, however, much of that investment has been focused on improving productivity and operational efficiency, often via point solutions tackling one step within a process.
Pioneers are pushing ahead with orchestration and connected workflows
A small cohort of leaders is already beginning to show what the next stage of AI maturity looks like. Much of that progress is being driven by agentic AI.
While the term has become something of a buzzword, what agentic enables is the significant part. It opens the door for applications that combine different sources of information, support multi-step reasoning and act more autonomously. In insurance, that means embedding AI decisions more consistently across individual workflows, rather than supporting isolated tasks.
Examples certainly exist of insurers taking this more expansive approach – around 30% of published AI use cases now connect multiple stages of the insurance workflow. Allianz’s claims platform coordinates several specialist AI agents that verify information, assess fraud risk, confirm policy coverage and support settlement decisions.
As these capabilities mature, they are also beginning to influence how insurers organise AI operationally. Connected workflows require shared platforms and closer collaboration between technology and business teams, moving AI further into the core operating model of the business.
Beyond AI efficiencies
While these further efficiency enhancements are important, and over 70% of disclosed use cases still focus on simple efficiency improvements, such as speed, cost reduction, and streamlining existing processes, the benchmark for success is beginning to move towards decision quality.
It is understandable that the sector has so far largely optimised for efficiencies, which are easier to measure and communicate, particularly while pressure to prove return on investment (ROI) builds in light of spiralling token consumption and costs. Faster document processing and shorter handling times produce clear before-and-after comparisons. Revenue growth or stronger underwriting performance, however, is harder to isolate and takes longer to demonstrate.
But as AI becomes more established and commoditised, efficiencies alone won’t be enough to set insurers apart. The real prize lies in improving the quality of insurance decisions.
This is partly due to the size of the opportunity. With claims accounting for between 60 and 80 percent of premium income, every improvement in pricing accuracy or fraud detection has the potential to compound across portfolios worth billions, giving it far greater financial impact than administrative savings.
Furthermore, while efficiency gains are valuable, they are also easier for competitors to replicate. Improvements in decision quality are where the early movers will create a durable and compounding competitive advantage. This is exactly where leading firms are focusing.
Challenges remain
Part of successfully deploying agentic AI, to connect workflows and improve decision quality, is making sure the right technical capabilities are in place. Changes to the composition of talent in the industry show that shift is underway. Data engineering is becoming a smaller share of AI hiring, with growth increasingly concentrated in AI development and software implementation. This suggests that many organisations are moving beyond building technical foundations, instead integrating AI into core business processes, creating the conditions for more connected applications to emerge.
Giving AI tools greater autonomy and influence also raises the bar for governance and control requirements, however. As AI becomes more widely embedded and the stakes get higher, insurers are strengthening the leadership and organisational structures that support AI. Eleven of the 30 insurers in this year’s Index now have a senior executive with explicit responsibility for AI, with two-thirds of those appointments made within the past year. While these roles extend beyond governance, they represent the growing importance of enterprise-wide ownership as AI becomes part of core business operations.
The leading insurers are increasingly treating governance as part of the product itself too. Looking again at Allianz, the company is building in traceability and compliance alongside its agentic claims capability, ensuring AI-generated decisions can be recorded, explained and audited as automation expands.
A vision for 2027
The signs are emerging that 2027 will be a pivotal year in the race to embed AI in insurance, where the gap between AI adoption and AI maturity becomes much clearer.
We are likely to see a wave of insurers talking about the return on investment they’re seeing from AI. Investors will be looking out for gains from productivity and operational efficiency. Customers and employees, meanwhile, are likely to place greater emphasis on how far AI leads to better outcomes and enhances service.
What’s going on behind the scenes over the next year is likely to be even more interesting than productivity, but harder to put a pound or dollar figure on right away, because demonstrating the value of better underwriting decisions or more accurate pricing takes longer to materialise.
The insurers that stand out over the next few years will be those laying the groundwork now, to leverage AI to improve the quality of decisions and build a durable advantage with this proprietary intelligence. Simultaneously, leading firms will not only continue to connect end-to-end processes across whole areas of the business, but will take the opportunity to actively redesign and optimise the operating model for an AI-enabled organisation.
Be the first to comment