Introduction

The increasing complexity of modern engineering systems has led to a surge in both the volume and granularity of requirements, making effective management critical for ensuring system quality, reliability, and compliance. To meet this challenge, Model-Based Systems Engineering (MBSE) and specifically, the model-based formalization of natural language requirements is becoming a cornerstone for driving efficiency and reducing risk across the development lifecycle.

Traditionally, creating use case diagrams, a foundational step in model-based requirements engineering, has been labor-intensive and time-consuming, particularly for large or evolving specifications. Today, AI-assisted modeling offers a transformative solution to accelerate this process.

By harnessing the power of Large Language Models (LLMs), organizations can partially automate the translation of natural language requirements into model-based representations. For real-world engineering use, these AI systems must process multiple requirements per use case and interpret heterogeneous, unstructured input data.

The MBSE Use Case Discovery Agent runs in the IBM Engineering AI Hub and is powered by IBM watsonx.ai. Together these components bring AI-driven automation directly into your workflow. The agent analyzes textual requirements to identify and recommend candidate actors and use cases, significantly reducing manual effort, improving consistency, and accelerating MBSE adoption as part of a broader digital transformation strategy.

Configuring AI Integration in IBM Rhapsody

Before You Begin

Ensure that you have a working workstation environment with Rhapsody 10.0.2 (or later). The AI integration brings intelligence directly into your modeling workflow, so preparation ensures a smooth setup.

Ensure that IBM Engineering AI Hub 1.1.0 (or later) is available, and you know the URL. See AI Hub Documentation

Step 1: Set Up the AI Environment

Download and extract the AIAgent.zip package.
Configure connectivity to your LLM (Large Language Model) server using one of the two simple options:

Option 1 – Modify rhapsody.ini:
Add the AI Hub URL under the [AIAgent] section.

Option 2 – Set an Environment Variable:
Define ENG_AI_HUB_URL in your system environment variables.

Step 2: Install the AI Agent Profile

Copy the AIAgent folder to your Rhapsody Profiles directory ($OMROOT/Profiles).
Open your Rhapsody project, then navigate to: File → Add Profile to Model → AIAgent.sbsx.

Once loaded, a new AI Agent menu appears in the toolbar.

Step 3: Discover Actors and Use Cases

In the model browser, select your requirements.
Right-click and choose AI Agent → Discover Actors & Use Cases.

Log in if prompted, and watch as the agent analyses requirements in real time.

The results appear in an interactive panel ready for review and refinement.

Step 4: Review and Create the Model

Inspect AI-generated actors and use cases, edit as needed, and confirm selections.
Click Create button.

Generate selected elements and diagrams.

Establish traceability links between model and requirements.
Display progress in the log window.

Step 5: Visualize Traceability

From the AIAgent profile, select Requirement Traceability Table Layout.

Create a new view and open it to see clear, automatically built links between requirements, actors, and use cases.
Select the package where you want to create the requirement traceability view.

Provide appropriate name to the traceability view and double-click it to open the view.

Once these steps are completed, you’re ready to begin using the MBSE Use Case Discovery Agent and experience the efficiency gains firsthand.

The MBSE Use Case Discovery Agent enables engineering teams to accelerate model creation, improve accuracy, and strengthen traceability all natively within IBM Rhapsody. It works with IBM Rhapsody Systems Engineering 1.6 (and later) too!, where the equivalent of the Rhapsody 10 AIAgent profile is built-in to the Rhapsody SE server.

By transforming natural language requirements into structured SysML models, it eliminates much of the manual effort and complexity traditionally involved in early-phase modeling.

More than a productivity tool, this capability serves as a strategic enabler helping organizations reduce modeling overhead, enhance collaboration across systems and software disciplines, and establish a robust foundation for digital thread continuity throughout the engineering lifecycle.