In the research field “AI-based predictive analytics”, we conduct research on methods and concepts of applying AI on digital predictive services. By doing so, we pay special attention to sustainable use of resources. The focus is on learning patterns and detecting anomalies in real-time data streams using machine learning. In this context, the topic of explainable AI plays an important role in ensuring the quality and interpretability of AI predictions. Our AI-based predictive services help to harmonize available capacities and expected demands - be it the utilization of parking spaces, the prediction of the occurrence of machine damage, or the prediction of peak consumption in energy management.