My name is Leon Starr and I have 20 years of hands on experience modeling the critical behavioral requirements of complex mission and safety critical systems.
In that time I’ve learned a lot about how MBSE can deliver powerful results and where it can fail dramatically.
You don’t want to experience that second part and I’m here to make sure you don’t.
(And if you do, I’m here to help you recover)
I work with a lot of modeling methodologies, tools, and standards, but there are a few universal principles that always essential to success.
Here are just a few that determine whether or not MBSE is providing genuine, lasting value on your project:
Reduce system complexity, not by hiding it in a nest of boxes-in-boxes but through detailed insights into the underlying domains.
Expose ambiguity, contradiction, and hidden gaps in even the most well written and managed requirements
Isolate and define platform invariant behavior so that your system works not just across platforms, but platform variations, and the continuous evolution of your implementation.
Uncover edge cases early so that they can be fixed rapidly and cheaply.
Provide transparency of key system behavior choices to both the engineers and the stakeholders.
Solve problems definitively so that you don’t end up with new hires rethinking and rediscovering the same problem throughout the project timeline.
Establish a clear, tangible path to both development and verification.
Models should lead directly to a stable framework without the need for large scale refactoring
And with regard to AI
AI helps build models and models help prompt AI
Models as human readable text are high fidelity prompts that lead more directly to safe, stable software — as opposed to natural language with all its faults.