Why model-based systems engineering stalls in so many companies, and the one sentence that tells a team when its model is done.
A model in an engineering organization can go wrong in two directions. Some teams model too little: they draw a few diagrams, nobody can answer a real engineering question with them, and the model is quietly abandoned. Other teams model too much: they capture every detail because they might need it someday, and the model grows faster than anyone can maintain it. Both end the same way, with a model nobody uses and a management team concluding that model-based systems engineering (MBSE) doesn’t work.
Researchers have asked practitioners what gets in the way. When the Systems Engineering Research Center surveyed 240 of them with INCOSE and NDIA for a 2020 report, the obstacles to MBSE adoption they reported most often were organizational culture, workforce knowledge and skills, and leadership support. MBSE tools were on the list as well.
Tim Weilkiens has been helping engineering teams with MBSE for 25 years, and the two directions are his description of a pattern he has seen “again and again, across domains.” He traces both to one cause. “Most MBSE adoptions stall for a reason that has nothing to do with the modeling language: nobody has clearly defined what the model is for,” he says. “And if you don’t know the purpose, you don’t know when the model is done.”
Tim is an executive board member of oose, the German consulting and training cooperative, where he and his colleagues train and advise engineering teams. He is a co-author of the specification for SysML, the Systems Modeling Language, and co-chair of the SysML v2.1 Revision Task Force at the Object Management Group (OMG), the standards organization that approved SysML v2 for final adoption in July 2025. “SysML v2 is the moment many companies reconsider how they engineer,” he says.
Enthusiasm, a Tool and a Notation Course
Tim doesn’t tell this as the story of one company. Client confidentiality rules that out, and he describes it as a pattern he has seen in many organizations across different domains. “A team starts with enthusiasm, a tool, and a notation course,” he says.
“What they don’t start with is a sentence like: ‘This model must let us decide X by date Y.’ Such a sentence is what tells you when to stop modeling.”
Tim Weilkiens, Executive Board Member, oose
The second reason he gives is the tool environment. If the tools don’t fit how engineers actually work, he says, every interaction with the model feels like overhead. During adoption, that’s a hurdle people have to climb over. After adoption he considers it worse, because “it becomes a permanently ineffective way of working.”
“People will tolerate a steep learning curve,” he says. “They won’t tolerate friction every single day.”
So when a team comes to him wanting to move from SysML v1 to v2, the language is not where he starts.
What He Asks First
“The first thing I do is not talk about SysML v2,” Tim says. “I ask what decisions the model should support over the next two years, and who will make them.” Then he and the team look at the system context: what the system interacts with, and who the stakeholders are. That, he says, is the starting point of SYSMOD, the Systems Modeling Toolbox, the MBSE methodology he edits.
“SYSMOD is a toolbox, not a process you have to follow step by step,” he says. “Teams pick the methods they need.”
Only then do the existing v1 models come out, and his advice is almost always not to convert them one-to-one. “SysML v2 is not a new version of the same diagrams,” he says. “It’s a language with a precise semantics, a textual notation, and a standardized API. Carry over what carries value and rebuild the rest in a way that uses what v2 can do.”
The next problem he describes arrives a few days after the training ends.
Learning on Monday, Stuck on Thursday
“Classic MBSE training has a flaw: the team learns on Monday and gets stuck on its own model the following Thursday, with nobody to ask,” Tim says.
In one consulting engagement, oose built the support around the team rather than in front of it. There was an initial training, then regular online Q&A, email support for quick questions, and focused workshops for deep dives. The team did the modeling work itself from day one, and the consultants kept it moving whenever it hit a wall. “The difference wasn’t more training,” he says. “It was a shorter distance between a question and an answer.”
oose has since turned that principle into the SysML v2 Learning Club, which combines self-paced microlearning, live topic sessions and a community for discussion. Tim describes it as a companion to the consulting engagements and a place for engineers to learn at their own pace.
His other example comes from a public course, and it involves an e-scooter fleet.
One Short Prompt
In a training on AI-assisted MBSE, the group worked with two AI agents: one that knows SysML v2, and one that coaches the SYSMOD method step by step. One participant generated an entire web-based simulation of his e-scooter fleet system from a single short prompt. Everything on the website, Tim wrote afterward, was reflected in the SysML v2 model, which captured the requirements, use cases and architecture.
“The point isn’t the AI. The point is that the model was precise enough for the AI to work with. A vague model gives you vague results, from humans and machines alike.”
Tim, Executive Board Member, oose
“AI can help you produce a lot, fast,” he wrote in a post about the seminar. “But the real question never goes away: Is the result correct?” His answer is to decide how you’ll test whether the AI result matches your intent before you prompt.
He knows the language from the inside.
Books as a Way of Learning
Tim has been at oose since November 2001 and active at OMG since 2002, where his profile lists him as a co-developer of SysML, UML and BPMN. He is still active in the ongoing work on SysML v2.
He also writes. His profile names Systems Engineering with SysML (Morgan Kaufmann, 2008) and Model-Based System Architecture (Wiley, 2015) among many books on modeling. “Writing books is my way of learning,” he wrote recently. “Behind the scenes, we have many discussions, discover modeling patterns and anti-patterns, and also identify improvements for SysML itself.” The four he counts as most important are The SysML v2 Book, Model-Based Product Line Engineering, Model-Based System Architecture and AI Assisted MBSE with SysML.
He founded MBSE4U, his own publishing house, “to put authors in the foreground and to publish work on MBSE quickly, while it’s still current.” It publishes the books on SYSMOD and on VAMOS, his method for variant modeling with SysML. He is co-founder and host of The MBSE Podcast, which he hosts together with Christian Muggeo, and the founder of X4P, a community for sustainability.
Asked for the one person who shaped how he works, he didn’t name one.
A Cooperative, and Five Days in 2020
“I’m not inspired by single persons,” Tim says. “I’m inspired by communities: the colleagues at oose, the SysML v2 team at OMG, the members of the Learning Club, the thousands of engineers who’ve sat in my trainings. Good ideas rarely come from one head. They come from people thinking together.”
oose is a cooperative that, in his words, “belongs equally to everyone who works in it, organized by consent instead of by direction from the top.” “It’s a luxury to do work that has meaning, rather than work you simply get through to earn money,” he says. “And meaningful work includes having a say in how it’s done.”
Consent, as he explains it, is not consensus. It doesn’t mean everyone has to agree, which “can take forever.” It means a proposal goes ahead unless someone has a reasoned objection. “The question shifts from ‘Is this the best possible solution?’ to ‘Is this good enough for now, and safe enough to try?’” he says.
The hard moment he chooses to tell is about that way of working. On a Sunday in 2020, the news announced that Hamburg would stop in-person events starting Monday. The oose campus was fully booked for the week. That week had to be canceled. The week after, oose was back, completely online. “In five days, we had sorted out everything: clients, participants, meeting technology, the legal side,” he says.
“Nobody gave orders. Everyone took what needed doing, because self-organization worked at its best exactly when it mattered most.”
Tim, Executive Board Member, oose
The project he calls closest to his heart takes the same idea outside the company.
A Learning Platform Owned by Its Users
haqoo, which Tim co-founded in 2021, “brings together the two things that have shaped my working life: what I learned about self-organization at oose, and what I learned from teaching engineers around the world.” He describes it in the future tense. It will be an online learning infrastructure owned by the learners, instructors, HR and anyone else who uses it.
“A lot changes when the goal isn’t to make as much money as possible, but to be financially stable and to offer as many services as possible, as affordably as possible, for learners and providers alike,” he says. “You can see it in the smallest decisions. In a meeting about a new feature, the question isn’t ‘Will this make us enough money?’ but ‘Can we finance it?’”
That, he says, opens the door to far more features, and it sets rules a profit-driven platform wouldn’t: learning providers can’t pay to rank higher in search results. “Learners find the best fit, not the biggest advertising budget.”
Where Tim Is Now
Tim co-chairs the SysML v2.1 Revision Task Force at OMG and is writing The SysML v2 Book with Vince Molnár. The book has been available as a pre-release, and its final release is set for January 15, 2027. At oose he trains and advises mostly at client sites around the world, and occasionally in open courses, among them a workshop for teams moving from SysML v1 to v2 and the seminar on AI-assisted MBSE.
He wants to hear from engineering leaders who are weighing MBSE, or who have already started and feel it stalling, and the conversation he hopes for is one with oose.
Outside work, he says: “I want to do something meaningful outside work, too. I’m the deputy mayor of my small village, and for more than 30 years I’ve been an active member of the Schutzstation Wattenmeer, a nature conservation organization that protects the Wadden Sea, the UNESCO World Heritage tidal flats on Germany’s North Sea coast. Both are, in a way, the same work as my job: a community deciding together how to take care of something that belongs to all of them.”
He used to run marathons, and he says that’s where he learned the rule he works by. “I don’t give up on an idea as long as I believe in it, even when everyone around me says it can’t work,” he says. What the marathons taught him is that “the moment you want to stop is rarely the moment you should.”
Asked what’s next, he gives one line: “Helping engineering organizations treat their models as a service others can build on.”
The sentence he wants to be remembered by is this one: “I help engineers think clearly enough that both people and machines can understand them.”
The Tim Weilkiens Playbook: A Model That Gets Used
- Write the purpose sentence first. “This model must let us decide X by date Y” is what tells a team when to stop modeling.
- Ask about decisions before you talk about the language. Which decisions should the model support over the next two years, and who will make them?
- Don’t convert old models one-to-one. Carry over what carries value and rebuild the rest in a way that uses what SysML v2 can do.
- Shorten the distance between a question and an answer. Put the support around the team, with Q&A, email and focused workshops, and let the team do the modeling itself from day one.
- Decide how you’ll test an AI result before you prompt. A vague model gives vague results, from humans and machines alike.
From the Diary of Tim Weilkiens
Tim Weilkiens is a Germany-based MBSE consultant, trainer and author and an Executive Board Member of oose.
| Role | Executive Board Member, oose. Founder, MBSE4U. |
| Based in | Greater Hamburg Area, Germany |
| Also | Co-Chair of the SysML v2.1 Revision Task Force at OMG |
“I help engineers think clearly enough that both people and machines can understand them.”
Editor’s Note
Executives Diary features Tim Weilkiens because his account of why MBSE stalls starts with a question any engineering leader can ask: what is the model for. He co-authors the language and still trains and advises the teams who use it. He answers a question about method with a sentence a team can write down, and a question about how a company should be run with five days in 2020. He keeps client names out of his examples, and so does this piece.
Executives Diary, Editorial



