Signal & Noise
Menu
AviationAI / MLSignal

A decade of predictive-maintenance AI, quietly preventing delays no one notices

easyJet Β· 2015–ongoing

What they did

Since a 2015 study with Airbus proved fault patterns were predictable from flight and maintenance data ahead of existing cockpit alerts, easyJet has built out a predictive-maintenance system (now 22 live models via Airbus's Skywise platform) that flags failing parts before they cause a delay or cancellation.

What happened

Between January 2019 and September 2025, easyJet reports the system helped avoid 171 major delays, 662 minor delays, and 1,343 cancellations. Despite a growing fleet over that period, easyJet has maintained aircraft availability at 98.63% and technical dispatch reliability at 99.57%.

The so-what

This is the least visible kind of AI win β€” nobody notices a delay that didn't happen β€” and it's also one of the most durable: a decade of steady, unglamorous model deployment against a single well-defined metric (part failure before it grounds a plane), with no viral moment in either direction.

Sources

Related case studies

← Back to all case studies