Customer Centric Development Dictates Siemens’ Current and Future Generative AI Roadmap

Siemens held an industrial-grade Artificial Intelligence (AI) event for analysts in Princeton, New Jersey that was a good showing of current capabilities and future deployment of generative AI, while providing insight into how much Siemens is investing in AI. For current deployments, Siemens Industrial Copilots, a set of virtual AI assistants available through Siemens’ applications such as TIA Portal and Teamcenter, was the main attraction, with use cases being added consistently and the value of the service growing. This aligns with Siemens’ internal thinking regarding generative AI, a practical tool to assist product and production engineers, as well as workers on a daily basis. Siemens already utilizes over 40 installments of generative AI across the company, with most obtaining value through employee efficiency. Siemens is taking a bullish approach to generative AI with a considerable chunk of Research and Development (R&D) investment planned for the next 12 months, along with an increase in hiring of AI specialists. With the build-out of use cases being the highest priority for Siemens, the underlying tone is that use cases must be pragmatic and have effective utilization by manufacturers.

Throughout the event, it was evident that Siemens is not looking for flashy industry-specific use cases, but rather is working on creating deployments of generative AI that provide tangible gains for manufacturers today. So far, Siemens has identified more than 400 uses cases for generative AI, many of which are currently available through Siemens Xcelerator and Industrial Copilot in Teamcenter. To identify and develop these use cases, Siemens speaks with customers to understand their pain points, and then works backwards to assess how generative AI can be used to rectify these concerns. This ensures that use cases brought into the Siemens ecosystem will be well received by customers on day one.

The most eye-catching use cases displayed at the industrial-grade AI event were for quality assurance and AI-assisted simulation. Quality assurance was shown through the Inspekto box (acquired in February 2024), a small-scale visual inspection aid for Small and Medium Enterprise (SME) manufacturers that has fixed costs of US$25,000 and variable costs of around US$10,000 based on image requirements. With generative AI-infused quality assurance, 100% of component defects can be found prior to product assembly, beating the average human, which can identify 60% of quality issues. To attain a defect detection rate of 100%, generative AI goes beyond the capabilities of general AI by generating synthetic versions of all potential component faults, built out from datasets of commonly seen issues. For simulation, generative AI has been used to speed up Finite Element Analysis (FEA) and Computational Fluid Dynamics (CFD) by consolidating vast quantities of data and forming new datasets with the goal of reducing the overall input workload for cloud-based simulation software. This significantly reduces the time needed to complete simulations, as the new working datasets are consolidated and refined to include only critical information vital to simulating what is required.

Throughout the event, it was evident that Siemens is not looking for flashy industry-specific use cases, but rather is working on creating deployments of generative AI that provide tangible gains for manufacturers today. So far, Siemens has identified more than 400 uses cases for generative AI, many of which are currently available through Siemens Xcelerator and Industrial Copilot in Teamcenter. To identify and develop these use cases, Siemens speaks with customers to understand their pain points, and then works backwards to assess how generative AI can be used to rectify these concerns. This ensures that use cases brought into the Siemens ecosystem will be well received by customers on day one.

The most eye-catching use cases displayed at the industrial-grade AI event were for quality assurance and AI-assisted simulation. Quality assurance was shown through the Inspekto box (acquired in February 2024), a small-scale visual inspection aid for Small and Medium Enterprise (SME) manufacturers that has fixed costs of US$25,000 and variable costs of around US$10,000 based on image requirements. With generative AI-infused quality assurance, 100% of component defects can be found prior to product assembly, beating the average human, which can identify 60% of quality issues. To attain a defect detection rate of 100%, generative AI goes beyond the capabilities of general AI by generating synthetic versions of all potential component faults, built out from datasets of commonly seen issues. For simulation, generative AI has been used to speed up Finite Element Analysis (FEA) and Computational Fluid Dynamics (CFD) by consolidating vast quantities of data and forming new datasets with the goal of reducing the overall input workload for cloud-based simulation software. This significantly reduces the time needed to complete simulations, as the new working datasets are consolidated and refined to include only critical information vital to simulating what is required.