AI and industrial design: how does it really work?
The integration of artificial intelligence in design and engineering processes is no longer an option, but a necessity to remain competitive. Leading companies are transforming traditional workflows without disrupting the existing infrastructure.
AI in the product lifecycle: who is winning?
Traditional PLM platforms are not replaced by AI, but enhanced through intelligent integrations that maintain control over critical data.
SAP has implemented Joule, an AI assistant based on natural language that operates as a universal interface for the entire enterprise ecosystem. Joule automates complex tasks: generates code for SAP BTP extensions, summarizes detailed financial reports, and helps users find information across S/4HANA, SuccessFactors, and Ariba.
- Cross-platform information search on SAP systems
- Automatic generation of code and documentation
- Summarization of complex financial reports in natural language
Data platforms like Snowflake, Databricks, and Palantir are redefining the approach to industrial data. Snowflake offers cloud-native data warehousing on AWS, Azure, and GCP, while Databricks provides a unified platform for data engineering and AI.
These solutions do not replace existing PLM systems, but represent an’augmentation of capabilities, not a disintermediation. PLM vendors are building barriers to’entry by rapidly integrating AI functionalities into their ecosystems.
From generative design to generative engineering: the leap in quality
Siemens leads the’evolution towards predictive systems that anticipate user’actions, going beyond simple geometry generation.
Siemens has moved beyond the concept of “Generative Design” to embrace “Generative Engineering”. The’goal is to automate the entire engineering process, not just the creation of shapes. The’approach is based on deterministic, transparent Industrial AI that complies with strict safety and precision requirements.
NX Command Prediction anticipates user’actions during design. Simcenter Neural Networks creates Reduced Order Models (ROMs) that accelerate simulations. Teamcenter integrates AI to manage the Digital Thread, analyze the’impact of changes, and identify duplicate parts in global systems.
| Features | Generative Design | Generative Engineering |
|---|---|---|
| Scope | Geometry and shapes | Entire engineering process |
| Predictability | Limited | Anticipates user actions |
| Integration | Specific tools | Complete Digital Thread |
| Models | Static | Dynamic ROMs |
The’Industrial Copilot developed with Microsoft represents a further step forward, combining cloud computing power with knowledge of specific industrial domains.
AI assistants in design: between code and natural language
AI assistants translate natural language requests into concrete technical actions, eliminating barriers between intention and implementation.
The real revolution lies in the’interface. Engineers no longer need to know complex syntax or navigate articulated menus: they describe what they want to achieve and the’AI translates into technical commands.
SAP’s Joule exemplifies this approach. A user can ask “create a Q1 expense report for the R&D department” and the system automatically generates queries, extracts data, and formats the document. The same logic applies to code generation for customizations.
Current systems use retrieval-augmented generation (RAG) to combine language models with domain-specific knowledge bases. This ensures accurate responses anchored to verified data.
Siemens NX Command Prediction goes further: it not only executes commands but learns from the user’s work patterns and proactively suggests next actions. This reduces design time and minimizes errors.
The’AI does not replace the designer, but amplifies their capabilities. The competitive advantage lies in the’intelligent integration between existing tools and new AI features. Companies that manage to maintain control over critical data while leveraging the’computational power of the’AI will achieve superior results.
Business models are changing. Pricing per user seat is giving way to usage-based or outcome-based models. When an AI agent can do the work of multiple people, paying per individual license loses meaning.
Evaluate today the’impact of an AI proof of concept in your design process. Start with a’circumscribed area, measure the results, and scale progressively.
article written with the help of artificial intelligence systems
Q&A
How are leading companies integrating artificial intelligence into existing PLM systems?
Not by replacing traditional platforms, but by enhancing them through intelligent integrations. SAP has introduced Joule, an AI assistant that operates as a universal interface, while platforms like Snowflake and Databricks increase data warehousing and data engineering capabilities without disintermediating existing systems.
What is the difference between Generative Design and Generative Engineering according to Siemens?
Generative Design is limited to the creation of geometries and shapes, while Generative Engineering aims to automate the entire engineering process. Siemens uses deterministic Industrial AI to anticipate user actions, as in NX Command Prediction, and to create dynamic Reduced Order Models that accelerate simulations.
What makes SAP's Joule a significant example of an AI assistant in industrial design?
Joule works as a universal natural language-based interface for the SAP ecosystem, allowing users to automate complex tasks. It can generate code for SAP BTP extensions, synthesize detailed financial reports, and search for information across systems such as S/4HANA, SuccessFactors, and Ariba.
How are AI assistants changing the interaction between engineers and design software?
They eliminate the need to know complex syntax or navigate complex menus, translating natural language requests into concrete technical commands. Siemens NX Command Prediction, for example, learns from user patterns and proactively suggests next actions, reducing time and errors.
How is the business model in the industrial software sector changing with the advent of AI?
Traditional per-user pricing is giving way to models based on usage or outcomes. Since an AI agent can perform the work of multiple people, paying for a single license no longer makes economic sense, pushing companies toward new evaluation metrics.
