According to the Association of British Insurers, fraudulent claims in the UK contribute an extra £50 to a policyholder’s coverage each year. In order to combat this, AND-E UK used a rule-based model to automatically flag and triage a subset of claims for further human investigation. Though this saved many hours, the number of cases sent to triage remained high, the model performance deteriorated over time, and the lack of explainability made it nearly impossible for investigators to budget their time and know what to investigate first.
In a highly competitive market, the impact of fraud on AND-E’s business added an unwanted £50 to the premiums of policyholders each year.
Mind Foundry worked closely with AND-E to understand the telltale signs of fraud from a human perspective and embed that into a three-pronged AI solution they could collaborate with to combat fraud more intelligently.
The solution incorporated more diverse datasets into a continuously learning model, significantly improving the quality of cases sent to triage and adapting to new fraud cases.
It prioritised flagged claims in an investigation dashboard, so AND-E could resolve them in order of the likelihood that they contained actual instances of fraud.
It accelerated the investigator’s ability to generate reports and close cases with natural language search options for structured and unstructured data, including hand-written notes and phone numbers.