At MODELS 2026, I had the opportunity to deliver a keynote titled "Your AI Agent Needs a Model", exploring how AI agents can interact with engineering models to support systems engineering activities.
The central message was simple: AI becomes more useful for engineering when it can understand and interact with structured models, rather than relying solely on documents and unstructured information.
Through live demonstrations using AI for Capella, I illustrated several practical applications:
- Architecture exploration and traceability: Navigating complex models, understanding dependencies, and identifying missing information.
- Model review and consistency checking: Detecting potential gaps and inconsistencies across system architectures.
- AI-assisted modeling: Creating and modifying model elements through natural language interactions.
- Methodological guidance: Helping engineers navigate modeling activities and apply engineering methods more effectively.
These capabilities are enabled by connecting Capella to AI agents through the Model Context Protocol (MCP), allowing them to interact directly with engineering models.
Of course, AI does not replace engineering expertise. Engineers remain responsible for architectural decisions, reviewing results, and validating changes.
Interested in seeing these examples in action?
I've shared more details about the keynote, along with the presentation slides and recorded demonstrations, on my personal blog: "Your AI Agent Needs a Model: Practical Examples from MODELS 2026"