AI is bringing radical change to how we develop systems. Models that used to take days now emerge from a single prompt. So the question is worth asking: will my MBSE skillset still be valuable in five years? A look at the current state of AI-assisted MBSE with SysML v2 suggests: yes - but only if we change along with it.
What Software Engineering Tells Us
Software engineering is usually a few steps ahead of systems engineering, which makes it a useful predictor of what's coming. Boris Cherny, one of the creators of Claude Code, has described in interviews just how radically his own work has changed. By December 2025 he hadn't written code by hand in 30 days, running several Claude sessions in parallel. By May 2026 he reported having hundreds of agents working day and night - and hit 150 pull requests in a single day. What's happening in software is a preview of what's coming for MBSE.
Three Shifts for Systems Engineers
From maker to judge: once SysML model generation is automated, our role shifts from creating to judging. Demonstrations make this shift tangible. A year ago, AI models still confused SysML v1 with v2. Today, trained models generate syntactically valid SysML v2 models out of the box - correct ISQ quantities, complete imports included. Entire SYSMOD walkthroughs, covering problem space, stakeholders, context, functional and physical architecture, plus views, can already be generated this way. Syntax and method are becoming less and less of a bottleneck - provided the AI knows your method and gets feedback through a correction loop.
From specialist to orchestrator: instead of focusing on one task, we'll increasingly coordinate multiple agents working in parallel on requirements, compliance, and architecture. The image of the systems engineer as an orchestra conductor isn't new - it's simply becoming literal. And AI today is locally brilliant but still a bit blind globally: the systemic perspective is still ours to bring.
Trustworthy AI: draft, not evidence: under emerging regulations, whatever AI produces is a proposal, not evidence. Only independent verification and validation create the assurance needed for certifiable products. That assurance comes from methodical engineering, not from AI engineering.
The Most Important Future Skill
Donella Meadows, systems thinker and co-author of "The Limits to Growth," described twelve leverage points for changing a system in her essay "Leverage Points." The highest one: the power to transcend paradigms - to change the mental model through which we see the world. That's exactly what's being asked of us now, as much of what we thought we knew about systems development ten years ago is being fundamentally rewritten.
Two Starting Points, One Restart
Companies without an MBSE history face a double transformation: they need to learn MBSE and how to work with AI at the same time. Companies with mature, established methods face a different challenge: unlearning ingrained ways of working and becoming more agile. The head start of the experienced group is real - but it's melting, because AI is accelerating the pace of change for both groups alike. The clock has essentially been reset.
Three Questions for Your Leadership
Do you have a documented, teachable method for what good systems engineering looks like - one that's also applicable to AI? Do you have a process that not only generates work, but also verifies and validates it? And: are you treating AI as a tool, or as a paradigm shift?
Asking these questions is itself already a step toward the most important skill of all: the power to transcend paradigms.
Do you have a topic that you would like us to write about? Let us know.
