Our Contribution: AI as a Methodical Extension of Proven Approaches

In their presentation, our consultants Jan Hecht and Yannic Labrot demonstrated how AI can be meaningfully applied in industrial processes today. The core message: AI in the form of advanced language models is not a replacement for established statistical or classical algorithmic methods. Instead, it serves as a powerful extension that ideally works hand in hand with these approaches—and in some cases even builds upon them. The type of AI that is suitable always depends on the specific task at hand and requires a clearly defined business objective.

We illustrated this through several practical examples:
AI based prediction models can reliably forecast mechanical material properties even before complex laboratory testing—an approach that does not replace existing metallurgical models but meaningfully enhances them. In quality analysis, too, the added value is evident: AI detects patterns in extensive measurement and test datasets far more quickly and helps derive causal relationships in a structured way. In other applications, AI models support production control by predicting process and throughput times, making bottlenecks visible at an earlier stage. Even in sequencing and production planning, AI acts as an additional building block that strengthens established rule based systems and improves the handling of complex planning decisions. And early generative AI applications in administrative environments—such as automated evaluation of technical documents or reports—already show how manual effort can be reduced.

Whether in production planning, quality assurance, or process optimization—AI unfolds its full value only when the operational purpose is clearly defined, enabling it to identify patterns, understand complex relationships, and support decision making.