Optimizing truck planning with AI

The context

A leading car manufacturer wanted to transform its spare parts logistics by moving from a traditional push model to a pull model. The goal was to streamline truck planning—a highly manual process involving millions of parts and 120 trucks shipped daily—while improving speed and accuracy across European operations.

The logic

To achieve this, the client needed an intelligent system capable of optimizing truck loading and leveraging historical data for better planning decisions. The solution had to integrate seamlessly with existing infrastructure and scale for production deployment.

The solution

We developed VTC (Virtual Truck Calculator), a tool composed of two key components:

• VTC Optimizer: Plans truck loading to maximize efficiency.
• VTC Learner: Uses historical data to extract insights such as nesting volume of parts.

The project was built in Python, with Snowflake for data storage, a FastAPI backend powering a React frontend, and infrastructure managed via Terraform on AWS and Azure. The predictive model, developed in Dataiku, was validated with the planning management team to ensure accuracy and usability.

Next steps in applied excellence

The full tool is nearly deployed to production, ready for end-users to integrate into their processes. Early results show significant improvements in time and accuracy, along with new opportunities for Kaizen-driven warehouse optimization.