Get in touch with Dr. Jürgen Krämer
Tell us briefly what it is about – we will pass your message on and make the introduction personally.
Podcast Episodes
Data Harmonisation with AIoT: What Do AI Agents Need for the Mapping?
For Krämer, a securely connected fleet is a precondition, not an achievement: without it, no AI project is worth starting. He separates two kinds of AI: classic machine learning for high-frequency machine data, where an LLM would blow up token costs, and agentic AI, which starts a root cause analysis when an anomaly appears. Both stand or fall on context: without asset relationships, units and documentation, a fault code like 4711 on a pump means nothing to an agent. Data harmonisation makes this tangible. In one plant the temperature sensor is called Hugo, in the next Willi. An Expert Agent reads the incoming data stream and proposes how to map it into the Cumulocity format – tested with OPC UA, Modbus, LoRaWAN, Protobuf, CBOR and Sparkplug. At Enercon, a rule-based remote action automates around 40 percent of turbine restarts; by Krämer's own account, no AI is involved. Autonomous intervention remains future work: prototypes yes, production no. Customers bring their own LLM, and Cumulocity trains no models on customer data, protecting the operators' domain knowledge. Krämer also warns against vibe coding in business-critical processes: AI-assisted development belongs at the application level, not in the system of record that carries governance and auditing. What you'll take away No AI project pays off without a securely connected fleet: bad data produces bad results. High-frequency machine data belongs with classic machine learning, not an LLM – token costs would spiral. The semantic layer is industry-specific, which is where domain partners come in. Data sovereignty means asking whether the vendor trains models on your operating data.
Scaling Industrial IoT: Cumulocity as Strategic Growth Partner
When companies attempt to build their own IoT platforms, scalability often fails due to complexity and high costs. In the podcast, Dr. Jürgen Krämer from Cumulocity explains how a platform-based approach overcomes these barriers – and why standardization is the key to efficiency. Cumulocity provides a scalable IoT solution that runs seamlessly on hyperscalers like AWS and Azure, helping companies become productive faster. Live data, device management, edge integration, and AI-driven analytics enable companies to shift maintenance from reactive to proactive – saving millions in service and downtime costs. Real-world examples like Enercon (30,000 connected wind turbines) and Flexco (real-time data for conveyor belts) demonstrate how businesses can boost efficiency and develop new digital services. If you want to scale your IoT projects successfully and make smart use of your data, you’ll find concrete answers here – straight from practice, for practice. Tune in now to learn how Cumulocity is taking industrial digitalization to the next level!
