Theodore Bergqvist, CEO and Co-founder of Turbotic, explores why so many of the insurance sector’s AI initiatives stall after the proof-of-concept stage and what carriers should do differently if they want AI to actually deliver ROI
Insurance firms have spent the last two years heavily investing and experimenting with AI; a stage that’s undoubtedly given many a much clearer view of exactly where the tech can deliver value.
However, despite AI being the top investment area for 3 out of 4 insurance CEOs, and insurers continuing to launch pilot after pilot, there is a clear disconnect between this activity and ROI, with concrete industry examples of meaningful financial impact or sustained operational change still few and far between.,
The problem isn’t a lack of AI – tools, pilots, MVP agents, automation history and vendor options are all in abundance – nor is it access to AI – data tells us almost all insurers already use it. The problem is a tendency, particularly in insurance, to stay in experimentation mode.
Most insurers understand that AI is something they need to address, but knowing where to begin is less straightforward. Where should they start, which processes are worth focusing on, and how will they know if it’s actually delivering value?
The real challenge now is moving out of this experimentation mode and into the next phase, where AI is governed, scalable and actually capable of delivering real, lasting business value.
What happens when AI is successfully scaled beyond experimentation
Having mentioned that examples of real ROI are relatively rare in insurance, other sectors are already showing exactly what can be unlocked when AI successfully scales beyond experimentation.
Take Evoke, for example, one of the world’s leading betting and gaming companies.
The beginning of its AI journey was a familiar story – fragmented automations across teams, no visibility into what was automated or where value was being created, multiple vendors and platforms with no unified strategy, lack of governance, standards, and centralized oversight, and difficulty measuring ROI or scaling automation beyond individual teams. But once a sustainable AI & Automation capability was built, the results were transformative, with £12 million in business value delivered.
While operating in an entirely different market, the gaming sector shares many similarities with insurance when it comes to AI barriers; from being highly regulated and increasingly margin-pressured to managing complex legacy systems, stringent compliance requirements and business-critical processes where mistakes carry significant financial and reputational consequences.
With this in mind, Evoke’s experience really does demonstrate that the path from AI experimentation to measurable value is less about technology alone and more about how organisations build the structures around it, no matter the sector. For insurers looking to make the same transition, several principles stand out:
- Learn to distinguish AI opportunity from AI hype: Fear of missing out is undoubtedly distorting AI investment decisions, both in insurance and across other industries, thanks to the sheer speed at which AI’s entered the boardroom. But it’s important to remember that while internal experimentation can be valuable, not every potential use case represents a genuine business opportunity and the ability to build an AI agent has been democratised, while building one that can scale has not. For insurers, this means resisting the pressure to deploy AI simply because competitors are doing so.
- Identify the use cases most likely to deliver measurable value: One way to avoid hype is to start with the business problem and understand where value can be created – not just experiment with technology hoping benefits emerge. In the P&C sector, for example, where as many as 6 in 10 insurers remain stuck in the exploration or proof-of-concept stage of AI adoption, only 35% explicitly link their AI strategy to business outcomes beyond efficiency, and more than 43% aren’t even tracking AI metrics. Only through thorough discovery, process analysis, feasibility assessment and ROI modelling, can we move away from unfocused experimentation and identify the processes where AI, agents and automation can deliver measurable operational impact.
- Understand that prioritisation matters more than technology selection: Most companies aren’t short of ambition when it comes to AI, but they do lack clarity, confidence and, crucially, prioritisation. What this means is that AI has created no shortage of ideas, but plenty of uncertainty about where to start. A business can quickly build a list of processes it thinks AI could improve; the harder question is which of those problems are actually worth solving. The temptation, of course, is to begin with the technology and work backwards, looking for somewhere to deploy it. But the organisations like Evoke that get real value from AI tend to start elsewhere – with a clear understanding of what needs to change and why.
- Assess operational readiness: A compelling, perfectly prioritised AI use case can only get you so far. Success also depends on the foundations supporting it. Legacy systems and fragmented data environments can create significant barriers when it comes to scaling AI, as can any weaknesses in processes, governance structures or teams, in terms of skills, ownership, process understanding and confidence. If we can take one lesson from a company like Evoke it’s that creating the operational conditions that allow you to deliver sustained value is vital. For many organisations, building those capabilities is a tough feat and requires a combination of internal ownership and external expertise.
Insurers are facing growing pressure to demonstrate progress with AI; boards are demanding action, competitors are announcing initiatives and technology vendors are making increasingly ambitious claims. But the winners won’t be the insurers with the most AI pilots under their belt; they will be the ones that know which problems are worth solving and have taken the time and care to build the capability to solve them.
Be the first to comment