Cutting Costs through AI-powered Claim Scanner

 Our client is a health insurer who was seeking to reduce fraud, waste and overspend in their claims process. To do so, they enhanced their customer app with a new AI-powered claims scanner. The tool could scan receipts, automatically populate claim forms and filter out ineligible claims.

The technology was powerful... but only if customers used it. Many people distrust AI or feel uncomfortable with automated scanning, and without strong adoption the tool would not deliver the intended savings. Customers also often submitted ineligible claims “just in case,” leading to wasted investigations and a poor customer experience when claims were later declined.

The insurer asked us to review the claims journey and identify behavioural solutions that would build trust, drive adoption and ensure the tool was aligned with the goals of the customer (e.g. clarity of knowing upfront what claims are/aren't covered, faster turnaround times) .

Through a thorough review of the app journey and customer research, we identified three key barriers to AI adoption:

• Habit: Customers were unfamiliar with the “AI scanning” and preferred the traditional process of entering details themselves.

• Distrust of AI:  Unless they are given a credible reason, customers intuitively do not like the idea of AI making a decision on whether their claim will be approved or not
   
• Uncertainty about eligibility: Customers weren’t sure what was covered, so often submitted claims that were destined to be declined.
   

These barriers meant that without the right framing, the new AI scanner could be seen by customers as something that is getting in the way of getting a claim paid. This is the very opposite of its intention: to help customers get eligible claims paid faster.

We designed behaviourally-informed communications and nudges across the app journey to build trust and confidence in the AI tool.  Key changes included:

• Expectation setting: Clearly explained the new process in simple terms – “upload your receipt and we’ll pre-fill your claim form” – to reduce confusion and give customers the right expectations from the start.
   
• Benefit framing instead of technology framing: We reduced the emphasis on “AI” and instead highlighted what mattered to customers: faster turnaround times and less hassle.
   
• Smart signposting and warnings: We introduced nudges when customers uploaded likely ineligible receipts (e.g. cosmetic procedures, till receipts), helping them avoid wasted effort. For example, if a customer tried to upload a cosmetic receipt, the app would display a clear message: “This looks like a cosmetic treatment, which isn’t covered by your current scheme.”
   

Our behavioural redesign has delivered measurable impact:

• €680,000 saved so far in 2025 – purely from reducing ineligible claims from being submitted and reducing wasted processing effort.
   
• Increased adoption and trust – customers are engaging with the tool because it’s positioned around speed and ease, not AI jargon.
   
• A smoother claims experience – customers now get instant clarity on what’s covered, avoiding the frustration of delayed rejections.
   

The next phase will expand the AI tool to more claim types, creating additional automation savings and speeding up claims approval even further.