Autonomous operations require more than accurate predictions: AI must be able to act safely within defined limits. At Future Digital Twin & AI Amsterdam 2026, Thomas Spatzier, Head of Platform Scaling & AI at Siemens Energy, describes the progression from connected assets and operations to intelligent systems and, eventually, autonomous action. He explains why that final step matters: a prediction informs a person’s decision, while an autonomous system can act directly on a physical asset.
For Siemens Energy, scaling AI means making it work across assets, use cases, vendors and operating environments. Thomas outlines the role of a common industrial AI platform in making this possible. Shared data meaning, reusable models and agent capabilities, consistent deployment and built-in governance can help teams develop solutions without starting again for every site or application.
The presentation also examines what makes AI trustworthy when it has authority to act. Thomas calls for specific permissions and operating boundaries for each agent, alongside oversight, independent controls, auditability and ways to recover or take back control. These safeguards need to extend through the software lifecycle, from checking the origins of data, code and models to automated testing and verified releases.
Finally, Thomas explains why deployment at the edge is part of autonomous system design. Systems acting locally must continue to behave safely if connectivity is lost. His central message is that scaling AI depends on engineering trust and control across the entire platform and lifecycle, as well as across the operations where it will be used.
Speaker
- Thomas Spatzier, Head of Platform Scaling & AI, Siemens Energy
