Practical Application of AI to MBSE

Model-Based Systems Engineering (MBSE) has become essential for managing complex system development, yet the steep learning curve and time-intensive nature of modeling tools like IBM Rhapsody often limit their effectiveness. This presentation introduces the pros and cons of applying artificial intelligence to Model-Based Systems and Software Engineering (MBSE). The presentation also includes a demonstration of its practical application using SodiusWillert AI Modeling Assistant (SAM) for IBM Rhapsody, a novel Model Context Protocol (MCP) server that bridges artificial intelligence with IBM Rhapsody, fundamentally transforming how engineers interact with such a modeling tool.

SAM enables AI assistants like Claude, Anthropic's AI assistant, but also all others like ChatGPT or even your own, to directly read, write, and manipulate Rhapsody models, creating unprecedented possibilities for automation and intelligent assistance in systems and software engineering workflows. By leveraging natural language interactions, engineers can now delegate complex modeling tasks that previously required extensive manual effort and deep tool expertise.

SAM provides comprehensive capabilities across key MBSE activities (some of which will be demonstrated), including:

• interpreting requirements documents and automatically generating corresponding models,
• analyzing existing architectures to summarize their purpose and identify issues,
• producing professional-grade documenting models and explanations,
• accelerating team onboarding through worked examples and educational support,
• reverse engineering of existing code into structured models, while also conducting thorough model and code reviews,
• performing requirements coverage analysis to identify gaps and conducts impact analysis for planned changes, helping teams understand cascading effects across complex systems,
• activating automations and leveraging existing models and documents for contextual understanding, making it adaptable to diverse engineering environments.

This demonstration illustrates how, by integrating AI into modeling, we can dramatically reduce the time and expertise barriers in MBSE, enabling engineers to focus on design decisions rather than tool mechanics, ultimately accelerating development cycles and improving model quality across the systems engineering domain.

What this presentation is about and why it matters

How can AI help with model based systems engineering without turning engineering work into a black box? Andy Lapping tackles that tension through a practical walkthrough centered on IBM Rhapsody and SodiusWillert’s SAM assistant. Rather than abstract claims, the session shows how an AI tool can read a model, compare it to requirements, add traceability, inspect implementation details, and support documentation and safety analysis in an interactive workflow. It also keeps the limits in view, especially context size, determinism, and the need for human review. This is a good fit if you work with Rhapsody, SysML, UML, or adjacent modeling workflows and want a grounded look at where AI can fit.

Who will benefit the most from this presentation

• Modeling engineers who already use IBM Rhapsody and want to understand AI-assisted workflows in practice.
• Systems engineers who manage requirements, traceability, and model consistency across changing designs.
• Team leads or architects evaluating whether AI can help with model review, documentation, or analysis.
• Engineers onboarding to UML, SysML, or Rhapsody who want a sense of how an assistant could shorten the learning curve.
• Practitioners concerned about whether AI outputs are trustworthy enough for engineering use.

What you need to know

Useful background for getting the most from the session:

• Basic familiarity with model based systems or software engineering concepts.
• Some awareness of requirements traceability and model reviews.
• Helpful, but not required, familiarity with IBM Rhapsody, UML, or SysML.
• A general understanding of what large language models and chat-based assistants do.

Glossary (terms used in this talk)

• Model-Based Systems Engineering (MBSE): An engineering approach that uses formal models as a primary artifact for defining, analyzing, and evolving systems. It helps connect requirements, design, behavior, and verification in a shared structure.
• UML (Unified Modeling Language): A standardized notation for describing software structure and behavior with diagrams such as classes, states, and interactions. It is often used to communicate and analyze design intent in model-based workflows.
• SysML (Systems Modeling Language): A modeling language derived from UML and adapted for systems engineering. It is used to represent requirements, structure, behavior, and relationships across hardware and software domains.
• MCP (Model Context Protocol): A protocol for connecting models, tools, and AI systems through a standardized interface. It lets an assistant access external context and perform actions through compatible integrations.
• Context window: The amount of text or structured information a language model can consider at one time. When the working set grows beyond that limit, earlier material may no longer be available to the model.
• Determinism: A property of a process that produces the same output for the same input. In engineering tools, determinism is often valued because it supports repeatability and easier validation.
• token: A small chunk of text used by large language models as the basic unit of input and output; typically a few characters long.
• LLM (large language model): A statistical machine learning model trained on large corpora of text to generate or analyze natural language outputs.

Toolbox (mentioned in this talk)

• PlantUML: A text-based diagramming tool for generating UML and other technical diagrams from plain text descriptions.
• IBM Rhapsody: A model-based engineering tool for systems and software design, including architecture, behavior, and traceability work. It is commonly used to manage models, generate code, and connect design artifacts to requirements.
• SodiusWillert SAM: An AI modeling assistant that connects to Rhapsody and can read, write, analyze, and help automate model work. It is designed to work through compatible AI backends and model integrations.
• Claude: A large language model and chat assistant used for general-purpose reasoning and text-based tasks. It can be integrated into workflows where natural-language analysis or generation is useful.
• ChatGPT: A chat-based large language model service used for natural-language interaction, drafting, analysis, and other text tasks. It is often used as a general-purpose AI assistant in tool-integrated workflows.
• IBM Bob: An IBM-branded AI assistant referenced as another possible backend for the workflow. It represents a model or assistant that can participate in compatible AI integrations.

Final thoughts

Practical, demo-driven, and candid, this session gives you a clear sense of how AI can sit inside an MBSE workflow without pretending that the hard parts disappear. The value is less in a list of features than in a usable mental model for where assistants help, where they need guardrails, and how they can fit into traceability, review, and documentation work. It will be especially useful for Rhapsody users and anyone weighing AI support for model-centric engineering. The result is a grounded look at a fast-moving idea, with both promise and friction left visible.

This overview is AI-generated from the session transcript. Spot an issue? Let us know.

Edited by the speaker on May 7, 2026.