AI in Planning: Where It’s Already Adding Value Today

An analysis by the AI working group within the ICV’s NextGen@Planning task force shows that AI can already be used productively today, particularly in standardized, data-driven, and repetitive processes. In contrast, it still faces limitations when it comes to strategic tasks and those that rely heavily on qualitative assessments.

The International Group of Controlling (IGC) Controlling Process Model 2.0 was used for the evaluation. Both the economic potential and the technological maturity were taken into account. A high level of maturity is particularly evident where sufficient high-quality data is available, processes are standardized, and results are reliably traceable.

The use of AI is particularly well-established in operational planning, forecasting, management reporting, and certain areas of cost, investment, and project controlling. Typical applications include automated forecasts, anomaly detection, plausibility checks, automated reporting, scenario analyses, and recommendations for action.

AI, on the other hand, is less mature when it comes to strategic issues, risk management, and, to some extent, business partnering. In these areas, qualitative factors, unique situations, complex interrelationships, and human experience play a greater role. Even in new or one-time projects, there is often an insufficient data set to make reliable predictions.

A real-world example illustrates how AI can already be meaningfully integrated into planning today: An AI assistant supports departments with cost center and project planning. Using natural language, it guides users through the planning process, analyzes historical data and current variances, identifies cost drivers and anomalies, and generates forecasts and recommendations for action. In addition, specialized AI agents can analyze data and, for example, detect anomalies or problematic project configurations. Planning scenarios can first be simulated in a sandbox and incorporated into the actual planning only after human review.

The key factor for success here is not technology alone, but rather the combination of data quality, process standardization, and business value. At the same time, obstacles remain: heterogeneous IT landscapes and data silos, insufficient data quality, a lack of process integration, insufficient AI expertise, acceptance issues, and questions regarding data protection, regulation, and explainability.

The key takeaway, therefore, is that AI does not replace the experience of controllers, but rather enhances it. Companies should not adopt AI with a “technology-first” approach, but rather start in a targeted manner where processes are data-driven, repeatable, and economically relevant. Where, on the other hand, context, creativity, and qualitative judgment are crucial, humans remain indispensable.