AI for Insurance Fraud Detection
The Challenge
Insurance fraud is difficult to detect: true fraud is rare, yet most AI systems flag too many false positives, wasting investigators’ time and eroding trust in the technology. Daisy Intelligence project addresses these challenges by introducing Mamdani fuzzy logic, a method that detects new fraud patterns without relying on historical data labels. Combined with reinforcement learning, the system can autonomously adapt to evolving threats. The solution integrates multiple analytics, including rule-based and supervised learning models, into a single, composite suspicion score that is both accurate and explainable. To improve investigator efficiency, Daisy will also develop a large language model (LLM) powered chatbot that explains alerts in plain language, replacing complex user interfaces with a streamlined conversational experience. These innovations aim to significantly reduce false positives and investigative time in group health and travel medical insurance.
“We’re proud to address current fraud insurance challenges by incorporating a reinforcement learning framework, which will autonomously enhance detection capabilities and reduce false positives. Additionally, our development of an LLM chatbot will be a game changer for raising efficiency for insurers.”
– Siva Vakalapudi, CEO, Daisy Intelligence Corporation
Investment
$
0.5
M
Scale AI investment
$
1.2
M
Total investment
Partners
