The World of Data 2025
Experience the data event of the year!
AI-based assistant enables intelligent fleet management thanks to large language modelling
As a leading telematics partner in Europe, idem telematics GmbH provides comprehensive telematics solutions for transport processes across various industries. Installed sensors collect data on drivers, cargo, and vehicles, making it accessible to customers through Cargofleet 3, Europe’s market-leading all-in-one platform. The result: platform users can continuously improve their profitability, customer satisfaction, and competitiveness.
To drive technological transformation in transport logistics, idem telematics approached b.telligent to enhance its telematics platform with an AI-powered assistant. This assistant would seamlessly integrate into the existing system and, acting as a virtual dispatcher, understand natural language inputs and autonomously provide data-driven responses to complex queries such as:
At the start of the project, the b.telligent team conducted a comprehensive analysis of the existing platform. Through in-depth discussions with idem telematics’ expert team and a dedicated workshop, the team not only examined platform functionalities and various data sources but also conducted careful requirements engineering to outline the project’s needs, challenges, and vision.
Based on these insights, b.telligent’s Data Science & AI experts designed a foundational architecture for the Large Language Model (LLM)-based assistant and implemented it using the b.telligent LLM Code Foundation.
A key challenge was managing the multiple data sources feeding into the telematics platform. To address this, the project was divided into specific milestones to ensure step-by-step progress and measurable success.
The first crucial step was connecting real-time data. The assistant follows a multi-stage process where LLMs navigate several steps to generate accurate responses. First, an LLM determines whether a data query is required. If so, it generates database queries to retrieve relevant live data. A second LLM then processes this data to generate a precise response.
The next significant milestone was integrating historical data. Thanks to its modular architecture, b.telligent was able to extend the assistant’s capabilities by incorporating additional processing steps. The LLM formulates database queries to retrieve historical data, allowing the assistant to distinguish between live and historical data queries effectively.
To evaluate the assistant’s performance, b.telligent developed a question catalog and conducted internal field tests. Based on the results, the team optimized the entire system in close collaboration with the client.
Throughout the project, b.telligent emphasized close cooperation with idem telematics. The overarching goal was to equip the client’s employees with the knowledge needed to use and further develop the assistant independently, eliminating reliance on b.telligent’s expertise after project completion.
AI-Concept
Workshops & requirement analysis to develop a concept
Architecture
Design of a multi-layered architecture based on LLMs
Data integration
Integration of various data sources into the assistant
Database queries
Automated database queries via LLMs
Knowledge transfer to the client team
Knowledge transfer through close collaboration with client teams
Customized LLM architecture: b.telligent developed an AI solution that understands natural language and automates complex data queries.
Seamless data integration: Live and historical data sources were intelligently connected and integrated into the telematics platform.
Efficient knowledge transfer: Close collaboration allows idem telematics to develop the AI agent independently.
Fast and precise responses to a wide range of user queries validated the prototype’s success, which was first showcased at IAA Mobility 2024 in Hanover. Previously, users had to manually visualize and analyze data to determine which vehicles or routes required attention. Now, they can interact with an AI assistant that promptly delivers precise answers to their inquiries.
Close collaboration and knowledge transfer enable the client to continuously optimize the assistant and bring the new product to market. Throughout the process, users maintain full control over their data—an essential factor in the increasingly interconnected logistics industry, ensured through collaboration with Microsoft.
Tune in to the podcast now!
Innovation and integration, as well as interoperability, are key factors of the Microsoft product development. Learn more about our collaboration!
Want a handy PDF version of our success story? Whether you need it for yourself or to introduce the project to your team, download it now and explore the full success story. Enjoy reading!
Did our success stories spark your interest? If you're facing similar challenges in data, analytics and AI and look for expert support, let’s talk. A brief call can reveal how we can help you move forward.