GO-JEK dominates the on-demand delivery and transportation market in Indonesia. With a click on a button, customers can order everything from food, groceries to massages.
Like Uber, Lyft and Grab, GO-JEK provides incentive programs for drivers that have achieved a high volume of trips. This prompts many drivers to take advantage of the system by creating fake driver profiles to simulate real trips.
Initially, their team took at least 30 minutes to detect bad behaviour, which would delay service to customers and allow culpable users to escape.
Consequently, they needed a comprehensive solution to help their team inspect trips and identify negative activity in real-time.
In collaboration with the GOJEK fraud and data science teams, Afi Labs built JARVIS to visualise errant behaviour and respond to changes in a constantly evolving landscape. Using machine learning, we create trip attributes to help classify these trips and ban drivers automatically.
With the help of JARVIS, GO-JEK no longer spends significant amounts of time filtering data. JARVIS seamlessly connects to their trip database and flags fraudulent behavior in seconds.