Home / Case studies / YS Transporte
Truck on the motorway (stock photo)
Case study · Transport

Dispatchers have
a gut feeling. AI has data.

How YS Transporte suddenly turned three years of telematics data into money without replacing their experienced dispatchers. With them.

Industry
Transport · Freight
Size
120 vehicles · 18 dispatchers
Duration
5 months · pilot to rollout
Services
AI consulting · integration

Good dispatchers. Poor tools.

YS Transporte had everything: experienced dispatchers with strong instincts, three years of telematics data, a modern TMS. And yet fuel consumption kept rising while on-time delivery fell. No one knew why. The data was there, but no one had the time to make sense of it.

The familiar question: "Do we need AI?" Our answer after discovery: "Yes, but not as a replacement for the dispatchers, as their new co-pilot."

Predicting instead of reacting.

We built a forecasting model from three years of historical data: expected driving times, risk spots for delays, optimal route combinations, fuel-consumption estimates per route. The model runs as a suggestion inside the existing TMS, the dispatchers still make the call. It shows how we put artificial intelligence to work — always tied to a concrete business process.

  • Forecasting model based on historical telematics and order data
  • Integration into the existing TMS, no tool switching for dispatchers
  • "What-if" simulator for route adjustments
  • Explainable suggestions: why the AI recommends what it does (no black box)
  • Weekly reports for management

Money on the road. On time at the customer.

After 8 months in production the numbers are clear, and the dispatchers are sold. Instead of the feared threat to their jobs, the AI has become a relief: less after-the-fact explaining, more time for the genuinely difficult cases.

  • −12% fuel costs on comparable routes
  • +18% on-time delivery at the customer's site
  • −30% empty runs through better route combination
  • Payback in 11 months (calculated without the image effect)

The AI suggests, the dispatchers decide — after eight months in production, the tool has become part of everyday work.

Pixelschnitzel · key takeaway from the project
Stack

The project's tech stack.

Python (modeling) scikit-learn XGBoost Spring Boot integration PostgreSQL REST API to the TMS Grafana reporting GDPR-compliant (DE)
Dig in.
Get in touch

30 minutes on the phone.
After that you know more.

Pick a slot or simply call. You tell us what is not working, we ask questions, and we tell you which process gets easier next. Free of charge, no contract.

Prefer writing? WhatsApp Email Contact form