Starting Point
The current reality of conflict in Ukraine marks a profound watershed both in the conduct of operations and in the way command posts are employed. Classic, concentrated Cold War-style command posts have become acute high-value, high-risk targets. The fusion of high-resolution intelligence, surveillance and reconnaissance with long-range precision munitions, coupled with the employment of long-range strike drones, has negated operational depth as a protective factor. In future, command and control will have to be exercised from much smaller and more mobile facilities. Protection through field fortification and mobility is becoming a dominant requirement.
In the Ukrainian theatre, a modular, decentralised command-post architecture now predominates, consisting of separated, buried yet rapidly relocatable command-post cells. In practical terms, this means that the same tasks must be completed in less time, with fewer personnel, and under more restrictive spatial conditions. At the same time, the required standard of output has remained equally high, because the quality of command and the success of operations is inseparably linked.
The recognised structural requirement to adapt the command organisation is transferable to any modern army and therefore also to the conduct of land operations in the Bundeswehr. The underlying concepts are currently being revised so that Land Forces capability development can be adapted rapidly. Since the distributed structure entails numerous limitations when compared with the traditional model, thought should be given at an early stage to how the emerging shortfalls can be offset. New capabilities in the field of generative artificial intelligence offer a wide range of possible approaches here.
Dispersed command-post cells, in which powerful, optimised AI models are operated locally on mobile servers and support the military decision-making process in a variety of ways, form the core of the conceptual approach presented in this article.
Evolution of Generative AI Systems
The civilian AI technology sector is currently undergoing a fundamental transformation. Until now, powerful generative AI systems could only be operated in large, cloud-based data centres. Considerable effort was invested in scaling these systems, which made their operation ever more demanding in terms of hardware requirements and energy consumption. Whether and when this trend will come to an end is difficult to assess.
More recently, it has become clear that smaller AI models are also delivering significantly better results and are gaining increasing acceptance. Unlike their cloud-based predecessors, they are often optimised for specific tasks. Their greatest advantage, however, is that they can be operated on local hardware, for example on local servers or powerful laptops. This trend, driven by innovation in software architecture and advances in hardware, creates the necessary precondition for building AI-enabled command-post structures.
Methods for reducing the size of AI models: Various methods are used to reduce further the size of AI models that are too large for local deployment. The key methods here are quantisation and the mixture-of-experts approach. In quantisation, the resolution of the model weights is reduced. The result is a significantly smaller model that nevertheless performs almost as well. Mixture-of-experts models use a high total number of parameters, but activate only a fraction of them for each query. This likewise reduces the required computational load. Both approaches can be combined very effectively, which reduces model size still further. Only this radical reduction in hardware requirements makes it possible to integrate high-performance AI capabilities organically into each individual, decentralised command-post cell without depending on a vulnerable cloud connection.
Operational Implications
Combined operations and combined arms warfare define the success of Land Forces in combat. Effects-based thinking and the ultimate synchronisation of all available cross-domain means, kinetic and non-kinetic, national and multinational, towards a common objective form the foundation of all action. Combined arms warfare aims at the precise orchestration of fires, manoeuvre and obstacles in order to exploit the enemy’s weaknesses and amplify the strengths of one’s own arms and services. Without this coordinated overall effect, the combat power of any large formation collapses.
This is precisely where the critical interface with command-post work lies. Staff work is the hub of synchronisation. On the basis of an integrated, cross-domain operational picture, it must ensure spatial and temporal coordination. In a distributed and reduced command-post architecture, this coordination requirement increases dramatically, while the time available to react decreases significantly. Sustaining a high operational tempo while incorporating all forces is a condition for survival. To master the complexity of this extraordinary human task in future combat as well, AI-enabled command-post work is shifting from an optional feature to an operational necessity.
The AI-Enabled Command Post: A Conceptual Approach
At the heart of the concept is a novel human-machine interaction based on generative AI and shaped by speech, text, and image and video information. It reduces the cognitive load on personnel at the command post by automating the processing and fusion of complex information flows, thereby preventing cognitive overload while improving decision quality. The way applications are used and data are accessed is changing fundamentally. The trend is moving away from time-consuming manual operation of applications and towards speech-based AI interaction with preconfigured AI agents.
These can be assembled via drag and drop and configured by means of a prompt. No programming skills are required. All that is needed is tactical expertise and linguistic understanding. Conceivable examples include AI agents for assessing the combat effectiveness of enemy elements, or for evaluating the enemy situation in sectors specified by the user. Command-post work can thus be further automated and significantly simplified. Over time, a repertoire of AI agents emerges that can be adapted as required and reused at any time.
To ensure the required functional scope, the central AI system must be coupled to the C2 system in use. The model context protocol, which is already widely used in the AI field, offers ideal conditions for this. The AI’s internal simulation models can then be fed ad hoc with real-time data. Time-consuming preconfiguration or manual data entry can be dispensed with. Identified courses of action can thus be simulated, analysed and assessed for their probability of success within minutes. The relevant command and tactical-employment principles required for this are extracted autonomously by the AI from the underlying body of regulations, orders and operation plans.
AI Functionality of a Tactical Operations Center
The need for rapid decision-making has long been a principle of tactics: “Whoever fires faster and hits better wins the firefight.” In the decision-making process, the corresponding maxim is this: the side that reaches the right decision fastest retains the initiative. In this context, AI provides the required acceleration by generating operational pictures within seconds, evaluating courses of action, and submitting recommendations to the commander.
Implementation of the concept is based on a system-agnostic approach that offers numerous advantages for integrating and processing a wide range of data sources. This architecture enables implementation close to open-source applications, thereby ensuring the flexibility and speed of innovation required in the spirit of Software Defined Defence. The aim is to provide an AI system with deep integration across all IT services, databases and applications employed at the command post, while remaining intuitive for the user to operate by simple means.
Implementation can proceed step by step, so that genuine added value can be generated early on even with limited functionality. The data access and chatbot-based interaction described above can be implemented directly without further development effort. Building on this, additional IT services can gradually be integrated, for example geospatial information services for automated terrain analysis, logistics databases for real-time resource queries, or sensor data streams for target identification, thereby progressively expanding the system’s functional scope.
Human Decision-Maker
Human judgement remains indispensable in every phase of the military decision-making process. The generative AI system acts as a catalyst, while the commander retains primacy: the commander confirms the AI-generated situational assessment, designates the main effort, and sets the evaluation criteria, such as casualty risk or speed of execution. This symbiosis is defined by two core areas:
Responsibility-led interaction: The design of the human-machine interface must ensure that context-based, sound human judgement and human control within a responsible chain of command are preserved at all times. To achieve this reliably, targeted adaptation of leader development is required.
Digital command competence: The successful implementation of AI-enabled applications requires command personnel to possess sound data and AI literacy. This includes a deep understanding of the analytical mode of operation of AI systems and of their contribution to decision-making, as well as the ability to evaluate AI-based tools critically and assess their methodological limits with precision. To secure this expertise on a sustainable basis, it must be established during initial training and deepened through continuous professional development. Only in this way can the necessary acceptance of operational AI employment be ensured and the technology’s transformative potential be fully exploited.
Summary and Conclusion
The need to disperse and reduce command posts is an inevitable and inescapable lesson from current conflict. To compensate for the resulting shortfalls in synchronisation and responsiveness, the integration of artificial intelligence is no longer an option but an operational necessity for maintaining effective command and control. The outlined concept of the AI-enabled command post addresses this challenge and identifies initial approaches to a solution, but it marks only the beginning of a transformation that will go much deeper.
As additional AI capabilities are increasingly integrated on the battlefield, AI-enabled command and control will progressively develop into AI-centric command and control. This is a process that is now beginning and must be pursued consistently. Given the pace of development, military capabilities and the underlying systems must be adapted continuously. This requires rigorous implementation of Software Defined Defence. Static, rigid architectures are obsolete. Instead, new functionalities must be built rapidly in prototype form and iteratively developed further in dedicated testbeds. Only in this way can resilient command competence in high-intensity combat be assured.
In summary, the concept of the AI-enabled command post posits a symbiosis of decentralised, scalable IT architectures with precisely defined AI-based human-machine interaction models. Through this interaction, command-and-control processes are significantly enhanced in terms of speed, precision and resilience, while the primacy of human decision-making is preserved at all times.