Henrik Schunk

Data Foundation for AI-Driven Production in Mid-Sized Manufacturing

The starting point is a sobering finding: most discrete manufacturing plants don't know in real time which machines are running and which are idle. Downtime causes end up handwritten on paper cards. Sorowka argues that technology hasn't been the problem for ten years — the bottleneck is the business follow-through: data-driven decision-making, transparency over performance, and a widespread not-invented-here mentality. With generative AI, the bottleneck shifts to IT/OT integration. Cybus itself positions Connectware, its own Factory Data Hub, as the answer to exactly this IT/OT integration challenge. Traub explains why data sovereignty becomes a competitive issue here: for many Hidden Champions, production know-how is the differentiating factor. Beyond the storage location, it's about the legal jurisdiction governing the contract and independence from any single AI model. It gets concrete in two places: Schunk describes a PCF playbook from a working group that allowed a mid-sized company to report the figure required by an automotive OEM within two hours. Sorowka presents the autonomous factory as a target vision, staged into autonomy levels 0 through 4, analogous to autonomous driving. Key takeaways If you don't know in real time which machine is idle, you don't know your bottleneck — and can hardly plan the ROI of a data project in advance. Generative AI delivers analysis ideas and prototypes within hours; the remaining bottleneck is connecting to real-world production. According to Traub, data sovereignty is a competitive issue, not a compliance one, because production know-how is the differentiating factor. A target vision built on autonomy levels makes investments justifiable, even when the ROI of an individual use case is missing.