Real-time dispatch with AI route planning

The challenge

Coordinators planned routes by hand on whiteboards and phone calls. Trucks ran half-empty on return legs, and customers had no idea when deliveries would arrive.

What we did

1. 01Built a live map of every vehicle, job and driver status with GPS updates every few seconds.
2. 02Trained a route optimization model on two years of delivery history.
3. 03Shipped a driver app with proof of delivery, offline mode and turn-by-turn handoff.
4. 04Moved the platform to autoscaling Kubernetes to handle seasonal peaks.