Introducing Alex: a digital employee making a real-world impact
Repairs, maintenance & tyres (RMT) is a prime enabler of delivering digital services at digital cost levels, and LeasePlan is hard at work to deliver exactly that.
“Since 2020, we’ve been developing a next-generation machine learning robot to automate maintenance approvals,” explains Job Visser, Artificial Intelligence (AI) Engineer. “Its name: Alex!” ‘
Using machine learning, Alex handles the bulk of straightforward maintenance requests, reducing costs for LeasePlan, saving time for RMT suppliers and leaving LeasePlanners free to focus on more specialised requests.
When a request for maintenance comes in, Alex assesses the likelihood of approval, with the process then automated according to thresholds defined by the business. “Above 95% likelihood, the request is automatically approved, up to a EUR 500 limit,” Job continues. “Under 5%, the request is automatically rejected. For anything in between, the request goes to a LeasePlanner for manual review – with the outcome being fed back to Alex, so that they automatically learn for next time.”
There’s a second string to Alex’s bow: life expectancy. “When requested, Alex will also determine when components like brake pads need to be replaced,” says Job. “Beyond that, we’re investigating ‘proactive maintenance’, where our AI model triggers the request rather than the driver or supplier.”
After a year-long pilot, Alex started automating approval decisions in Austria and Greece in 2022. By the end of August, 13% of all maintenance jobs in Austria and 52% in Greece had been automatically approved. Alex is now running in eight other entities, with LeasePlanners using its predictions as expert advice.
“We’ll soon flick the automation switch for Alex in other countries, too,” Job says. “It’s really exciting: after some onboarding – much like a human employee! – the model can now train itself. Alex is already making a powerful impact on LeasePlan’s business efficiency, not to mention helping our RMT suppliers and colleagues. Best of all, this is just an early step towards fulfilling the potential of machine learning for our entire fleet operations.