1. When AI becomes a collaborator rather than a tool
Applications of artificial intelligence (AI) have long arrived in logistics. Yet solutions that address isolated sub-problems still dominate. However, this hardly does justice to logistics’ true strength – its ability to network and control highly complex processes. “The core competence of logistics lies in coordinating many individual steps in such a way that highly complex tasks can be reliably accomplished”, says Andreas Nettsträter, CEO of the Open Logistics Foundation. “Now the question is how artificial intelligence can provide support at this central point”.
The often-expressed hope that artificial intelligence will solve all logistical questions in the form of a universal system – in other words, a “ChatGPT for logistics” – will certainly not be fulfilled. Instead, a different scenario is emerging: an ecosystem of specialised AI agents orchestrated to manage supply chains. AI agents typically operate within a clearly defined area of responsibility, react to events, and trigger subsequent processes in a targeted manner. The particular added value for logistics lies in their interplay: when multiple specialised AI agents act together, an end-to-end process along the entire value chain becomes possible.
For AI agents to be integrated into logistics processes and usable along the entire value chain, the underlying workflows, data models, and interfaces must be comparable and defined in a binding manner. “AI delivers its benefits above all, where structured and comparable data are available”, Andreas Nettsträter sums his thoughts up.
2. How data models become “AI-ready”
Within the Open Logistics Foundation, companies have already developed various components and data models that can be specifically used for the development and deployment of AI agents.
- One example is the electronic consignment note (eCMR). The solution developed by the member companies of the Open Logistics Foundation – the so-called OLF-eCMR – is based on structured, standardised data. This makes documents directly machine-readable and easier for AI to process. It enables automatic extraction and validation of content, handover to downstream processes, and integration with transport, billing or compliance systems. In addition, the three status messages of the eCMR – “Provision by Sender”, “Loading by Driver”, and “Acceptance by Recipient” – are legally effective through the respective signature and act as triggers for subsequent processes, such as initiating automated invoicing or payment release.
- Another example is the Tracking Event Model of the Open Logistics Foundation. Companies have agreed on a standardised description of status events along the transport chain. The model thus provides an ideal basis for AI agents that can interpret events in real time and automatically derive subsequent actions. For example, an agent can evaluate incoming status messages and automatically trigger notifications, suggest re-routings, or re-coordinate time windows.
- Currently, companies in the Open Logistics Foundation are also developing a basic model for timeslot management. This could allow AI agents to dynamically manage time slots based on real-time data and capacity information. Instead of static bookings, an adaptive system emerges that can respond to delays, utilisation, or changed priorities and replan or optimise time slots accordingly. Through the use of AI, the examples mentioned can be fully automated. Simple agents can accelerate usage while simultaneously simplifying the dissemination of standardised data models and status events.
3. Why open source is an enabler for agentic AI
The current shift towards agentic AI, which acts proactively and “works on” tasks, is not only a technological but also an organisational question. Agentic AI should therefore not be conceived as a centrally provided solution, but rather as the result of the interplay of open components and company-specific integration. “Companies will only deploy AI agents in critical end-to-end processes if they can understand the decisions, retain control over their data, and integrate systems securely”, identifies Jens Leveling, Technology Advisor at the Open Logistics Foundation, citing two main reasons for open source: traceability and trust in open implementations. Open architectures and open source approaches enable the connection of different systems – a fundamental prerequisite for seamless value chains. Jointly developed components also reduce implementation effort and accelerate innovation. In addition, companies avoid dependencies on proprietary platforms and retain control over their processes.
In principle, the Open Logistics Foundation develops only basic functionalities, so-called commodities. Agentic AI and the underlying data models are not, in themselves, elements that differentiate competitively. Therefore, it makes sense for companies to develop them together and provide them as open source. Each company thus has the opportunity not only to use them but also to reuse them commercially. Andreas Nettsträter says, “The same principle also applies in the field of artificial intelligence: the actual differentiation in logistics does not arise in the AI agents themselves, but in the orchestration of AI agents within concrete application environments”. For companies, this opens the opportunity to develop their own solutions based on these building blocks and deliberately leverage the orchestration of AI agents as a competitive advantage.
