FAICE addresses the integration of generative AI into collaborative engineering workflows. The initiative focuses on two primary use cases: a SysML Assistant embedded in CATIA Magic for MBSE® and a Standards Chatbot for retrieving licensed standards content.
FAICE aims to overcome systemic obstacles rooted in established models of data access, control, and licensing. The project ensures that AI responses are trustworthy and grounded in verifiable source material.
For the SysML Assistant, the goal is to support modeling tasks inside an industrial tool environment and facilitate collaboration between partners.
For the Standards Chatbot, the focus is on enabling secure, legally compliant, and granular access to sensitive or licensed data across corporate boundaries.
Added Value: FAICE offers practical approaches for the secure and trustworthy use of generative AI in engineering across company boundaries.
Objectives: The project aims to improve collaboration by establishing a secure and industry-compliant framework.
Interim Assessment: The project has successfully demonstrated a tangible, working prototype for both the SysML Assistant and the Standards Chatbot. Findings indicate that core technical components are mature, data sovereignty can be maintained, but adoption depends on the evolution of commercial and operational landscapes.
The project delivered a working SysML Assistant that converts natural language into SysML models. Users reported efficiency gains in early modeling tasks, such as creating structure and behavior diagrams. Key user requirements identified include persistent context across sessions and dialog-based clarification to prevent errors.
For the Standards Chatbot, users can now interact with a virtual assistant to retrieve specific content from standards based on individual licenses. The system successfully enforces granular permissions, ensuring users only see information they are authorized to access.
For service providers, the Standards Use Case defined the role of an AI Orchestrator acting as a neutral intermediary. The architecture is based on a "zero-knowledge" principle where the orchestrator routes queries without retaining sensitive data. FAICE identified the Catena-X data space as a potentially suitable ecosystem for deployment, enabling policy enforcement. This allows providers to offer "Standards-as-a-Service" where data is accessed via secure Interfaces (APIs/A2A/MCP) rather than static document downloads.
FAICE has determined that conventional licensing models (per user/per document) must be critically reviewed before the use of AI agents that paraphrase and quote, for example. The project recommends a transition to "AI-first" business models. These include Pay-per-API call/Pay-per-Token models and Project-based subscriptions that grant partners access to specific standards for a project's duration. A "Freemium" model was proposed to make metadata freely discoverable by AI agents, driving transactions for full content access.
Technical: A major challenge was “data readiness,” as many standards (PDF, ReqIF, SysML) are designed for exchange between organizations and tools. However, this does not imply that AI models can also work sufficiently well with these data formats. This is exacerbated by the fact that, in some cases, metamodels allow for individual interpretations of the standards, and, for example, in cross-company collaboration, SysML profile dialects and incompatibilities of abstraction levels are a further challenge.
Organizational: The primary barrier for the Standards Chatbot is the "Data and Licensing Bottleneck". Existing licensing models do not support the granular, on-demand access required by AI agents.
For the SysML Assistant, access to experienced SysML practioners remains critical for the goal to address real-world collaboration scenarios.
In 2026, FAICE will focus on "Technical Extensions of the SysML Assistant" to address real-world collaboration challenges.
1. Cross-enterprise model exchange: The assistant will detect deviations in company specific profiles and suggest or perform automated repairs.
2. Support for multiple versions: Addressing differences in SysML versions (1.x/2) between partners.
3. Integration of additional modalities: Practical application of "Smart Standards" (IDiS) and ReqIF to improve the assistant's efficiency and reliability.
The release of SysMLv2 offers an opportunity to investigate the Assistant capabilities to support v1x-to-v2 translations.
The project successfully built working prototypes, proving the viability of secure, compliant frameworks addressing real day-to-day issues in the industry. These findings are published in 2 whitepapers, a recommendation and a webinar.
A DIN/DKE Project (“IDiS”) uses the results already in progressing international discussions to push for machine-digestible standards and norms.
Markus Franke (Schaeffler, Project Chair): “With FAICE, we were able to put our finger on the sore spot. The experiences we gathered reveal that there is still a long way to go to enable collaboration without deliberate obstacles. Openness often means changing everything we were used to doing and were willing to share.”
Michael Maletz (AVL): “AI is set to transform the way engineering is done. Through the findings of this project, the vision for applying AI within a heterogeneous MBSE environment is becoming reality — we are shaping that future. FAICE offers a glimpse into how engineers will benefit from AI and lays the foundation for organizations to prepare for the workplace of tomorrow.”
Michael Haller (ZF): “FAICE demonstrates how AI can simplify MBSE and bridge gaps in collaborative engineering. By transforming natural language into structured models and seamlessly integrating work across organizations, formats, and SysML versions, FAICE frees engineers to focus on true architecture and innovation rather than tool complexity. This marks a major step toward a future where AI becomes an integral partner in engineering workflows”
2024/5: Markus Franke (Schaeffler)
2026: Michael Maletz (AVL) / Michael Haller (ZF)
Participants
We are eager to help you with any questions concerning your project or selection of publications
TELEPHONE HOTLINE
+49 6151 9287-336
For more inspiration, news and information about prostep ivip.
