#EXL

EXL's graph model finds fraud in the edges

EXL showed on AI Engineer how adding graph context lifted fraud capture 27-33% at fixed alert rates and held P99 latency to 3.1ms.