How Data, AI, and Agents are Changing Supply Chain Planning
This article is a second part of a six-part series currently appearing in DVZ (Deutsche Verkehrs Zeitung, dvz.de), analysing the future of each station in the supply chain. Together with SAP Prof. Hoberg and his team spend the last year studying how supply chains will realistically develop and included input from 660 supply chain experts.
In the discussion about the future of supply chain management, one area is increasingly at the center: planning. Supply chain planning not only affects operational efficiency, but directly influences a company’s service quality, fixed capital and responsiveness. Our joint study from SAP Business Consulting and Kühne Logistics University clearly shows that companies that are already planning proactively today can avoid bottlenecks, optimize costs, and exploit potential in a targeted manner.
Supply chain planning is the central lever for anticipating problems before they arise. Companies that anticipate bottlenecks and proactively manage capacity, inventory and resources avoid hectic “fire fighting” in day-to-day operations. Planning thus acts as an early warning system that not only enables stable service, but also more efficient use of resources and proactive action. Effective planning also requires seamless integration and evaluation of a variety of internal and external data sources – from granular sales data to basic economic data, to production, procurement, and distribution plans. Different planning levels and horizons need to be connected, and this is where there are huge opportunities for artificial intelligence that can process data volumes, identify patterns, and improve predictions. Advanced planning processes will become a matter of course over the next ten years. If you do not achieve at least a modern, integrated planning level by 2035, you risk not only a competitive disadvantage, but also the loss of operational control capability.
Our survey of supply chain decision-makers (see figure) provides a clear picture of the future. Many expect planning to be much simpler and more intuitive in 2035 than today. More than half expect generative AI and GPT models to have a significant impact on planning processes, and much of it sees data quality as a critical factor in planning decisions. In addition, many expect the breakthrough of new AI-powered forecasting tools that incorporate additional data sources and significantly improve demand planning. Around a quarter expect advanced risk analysis to provide the ability to proactively detect and mitigate disruptions.
However, the status quo shows how big the gap between aspirations and reality is. Only a small number of companies rated their planning as mature in our survey. Similarly, few are actively experimenting with GenAI or GPT applications in the planning environment. Only a minority has very accurate planning data harmonized from different sources. Highly automated demand planning with minimal human intervention is the exception, as are planning systems that provide clear recommendations for action in the event of disruptions. This discrepancy shows that the path to truly modern, intelligent planning is still long and requires targeted investment, clear priorities and structured action.
In order to close this gap, different approaches must be pursued. Companies where planning is a strategic core process or requires very individual processes can realize enormous potential with agent-based AI. With the advent of new agency tools, most repetitive, error-prone work is automated, allowing planners to focus on value-added analytics and decisions. Companies with standardized processes and a low strategic relevance of planning will be able to procure their planning function more as a service in the future. Technology and platform models could enable true end-to-end planning across company boundaries for the first time. Simulations and digital twins will also play a growing role. Although these often fail today due to complexity, modeling effort, and scalability, AI-powered modeling agents could significantly accelerate and improve the accuracy of simulation creation and updates in the future, making their use more practical in larger networks.
Regardless of the individual target image, there are measures that every company should tackle immediately (“no-regret moves”). This includes building robust data governance and harmonizing master data as the foundation for any data-driven planning. Unfortunately, the principle of “garbage in, garbage out” also applies with AI: without high-quality data, even the best AI system cannot deliver meaningful value. A consistent, company-wide IBP system enabling a S&OP process provides the basis for a common planning goal. Initial pilot projects with agent-based sales forecasting help gather experience with AI-powered planning approaches. Finally, it is important to actively link existing planning platforms, such as SAP IBP, with AI services to unlock integration potential early on.
Supply chain planning is now more important than ever and will become a key differentiator between successful companies and those left behind by 2035. The technologies to make planning smarter, faster, and more resilient already exist – but they must be combined with clear goals, a resilient data foundation, and organizational readiness for change. Those who lay the foundations today will not only be able to plan faster and more precisely tomorrow but can also act strategically before others can even react.
This article was written in collaboration with Prof. Kai Hoberg (Kühne Logistics University) in the context of the SAP-KLU study "The Supply Chain of the Future”.
