The shortage of skilled workers and competitive pressure are making themselves felt in everyday working life. Cooperation between man and machine can open up completely new possibilities here. Assistance systems based on generative AI can, for example, generate code and support operators with practical instructions for technical tasks. In this way, generative AI can change the industry and unleash potential for efficiency and sustainability.

The Siemens Industrial Copilot is an assistant supported by generative AI. Its core lies in the intelligent use of process and machine data, which can be recorded efficiently and in a standardized manner using the OPC UA interface. This technology ensures that operational data can be seamlessly transferred and analyzed in real time, helping operators to respond more quickly to different machine statuses and potential problems. At the same time, the co-pilot makes it possible to provide important documents (manuals or standard operating procedures) directly on the machine. This digital document repository supports operators with maintenance and problem solving. AI can also provide contextual information based on machine data and documents, which increases safety and efficiency.

© Siemens

Assistance systems based on generative AI, for example, can generate code.

By using state-of-the-art technologies such as AI and real-time data analysis, machines and processes are becoming smarter, more sustainable and more user-friendly. Examples of applications for the co-pilot include:

Energy Advisor: Detects anomalies within the energy data of production facilities and automatically identifies measures that offer potential savings. „The operations team is supported with optimized energy reports and shown how energy consumption can be sustainably reduced, which saves costs and protects the environment in the long term,“ says Daniela Borgmann (Project Lead R&D).

Shift and Operations Assistant: The interactive, AI-based Operations Assistant supports the operations team by acting like a personal data scientist. It provides access to all relevant information during shifts, provides user-friendly KPI summaries and reports and minimizes the effort involved in shift changes. In addition, the assistant correlates the behavior of different machines and production lines in order to gain deeper insights and increase operational reliability. „We integrate innovative language models into industrial systems. This allows the operations team to receive advice on optimization via chat,“ says Thomas Blumauer-Hießl, Researcher at Siemens Austria.

Large language models (LLMs) can „hallucinate“, i.e. generate plausible-sounding but false information. Retrieval Augmented Generation (RAG) helps by incorporating external knowledge from trusted sources, which improves accuracy. The fine-tuning of the LLM optimizes it specifically for industrial requirements, which increases the relevance and reliability of the answers in operation.

„Siemens offers a powerful GPU cluster (computer cluster for very fast calculations) on site. This allows AI models to be adapted to company-specific requirements quickly and in compliance with data protection regulations,“ explains Daniel Schall, Head of the Distributed AI Systems research group at Siemens Austria. Companies benefit from increased data security and the ability to efficiently adapt AI applications to their specific needs.

Siemens offers a powerful GPU cluster on site. This means that AI models can be adapted to company-specific requirements quickly and in compliance with data protection regulations.

Daniel Schall, Head of the Distributed AI Systems research group, Siemens Austria

© Siemens

Despite the enormous potential of the Industrial Copilot, companies often face hurdles when integrating it into existing systems (control systems, control, MES, PLM). Siemens solves this problem – standardized integration blueprints enable the simple implementation and scaling of copilot solutions (e.g. through integration into control systems such as SIMATIC PCS7, PCS neo, Siemens OpCenter Execution, etc.).

Efficient deployment of Siemens Industrial Copilot on an edge device enables seamless integration into automation systems without the data having to leave the store floor. This protects privacy and enables unrestricted operation without network outages and Internet connection problems.

The Siemens Industrial Copilot uses real-time data and relevant documents to support operators, minimize errors and optimize machine operation. Adaptive guidance improves precision and efficiency, while the co-pilot learns from interactions and continuously optimizes recommendations. Fine-tuning with company-specific data makes the co-pilot more precise and it can provide support through customized data sources and regular performance checks.