The End of The Blank Page: How AI Helps Systems Engineers Get Started?

Conversation with our experts #1

For many systems engineers, one of the hardest parts of model-based systems engineering (MBSE) is simply getting started. Beginning with a blank page means hours of manually entering requirements, creating elements, and integrating them into a consistent model. It is tedious, resource-heavy work and a real challenge to productivity.  Artificial Intelligence is beginning to change that reality. Instead of starting from scratch, engineers can now rely on AI to generate a first draft of a model from requirements, analyze an existing design, or even reverse-engineer it into something understandable. AI does not replace engineering expertise, but it can relieve the burden of repetitive tasks, surface insights more quickly, and help engineers concentrate on the core aspects of system design.  

To explore this shift, we spoke with two SodiusWillert-recognized Rhapsody gurus and IBM champions: Walter van der Heiden and Andy Lapping. Andy and Walter are also the creators of AI Modeling Assistant for IBM Rhapsody, a SodiusWillert project that began as a casual brainstorm and is about to become a very promising solution.

AI Modeling Assistant for IBM Rhapsody is still evolving, but its early development illustrates an important point: AI is no longer a distant vision in MBSE. It’s already demonstrating how it can act as a virtual and reliable assistant for engineers within their workflows in tools such as IBM Rhapsody, enabling faster progress while preserving engineering standards.

Enjoy this first episode of our series of interviews with SodiusWillert experts!

1) What’s the biggest driver in implementing AI in Systems Engineering, and more specifically, in MBSE? 

Andy: AI has become unavoidable. It’s everywhere, and every industry is finding a way to use it. The good news is that for model-based systems and software engineering, it isn’t just a trend. AI has clear and valuable applications. The main driver is that AI is now accessible, and engineers can finally use it to address their long-standing challenges. The problems in MBSE haven’t changed, but AI gives us new tools and new possibilities to solve them.  

Walter: Engineering teams can’t afford to ignore AI. It relieves a significant burden from engineers by automating repetitive or complex tasks and acting as a strong support system in their daily work.  

Andy: Exactly. Think about the possibilities: an AI assistant can help someone learn by generating a model as an example, review an existing model to identify issues and suggest fixes, or even create a first draft of a model from a set of requirements.  These are just a few cases; there are countless ways AI can support MBSE.   

2) How can AI best support engineers in becoming more productive while ensuring that efficiency goes hand in hand with the rigor and quality required in MBSE? 

Andy: For me, the biggest gain is speed. I might know exactly what I want to put in a model, but doing it manually slows me down. AI can add content much faster and can also process information quickly. For example, in IBM Rhapsody, there are thousands of properties that control how it works. I might know a property exists that can solve a particular problem, but can’t recall exactly what that property was called. AI can read the documentation, understand my intent, and point me to the right setting immediately – or even set the property for me.  

Walter: Yes, exactly! There are thousands of properties across different levels, and finding the right one can take time. AI lifts that burden. More importantly, it allows engineers to focus on the actual problem they’re solving, not on navigating their tool. An engineer designing a controller for a car or an aircraft wants to think about the solution, not spend mental energy clicking through menus.  You can ask AI to make this for you, and it handles the repetitive steps. That frees your mind to focus on the design challenge. Of course, it’s not foolproof; you still need to review and think critically, but it creates the mental spaces to do that effectively.  

 Andy: And it’s important to remember that AI is just a tool like any other tool. Owning a saw doesn’t make you a carpenter, right? It’s the same thing here. AI helps, but it still needs a skilled operator. Think of it as the difference between using a screwdriver to make a hole in a brick wall versus using a drill. The drill is faster and more effective, but you still need to know how to use it correctly if you don’t want to lose a finger or hire a plasterer to fix your mistakes.  

Walter: That’s a key point indeed. AI doesn't replace engineers; it supports them. It helps them stay focused on rigor and quality by removing distractions. 

Andy: And I would add one last thing about speed: how quickly AI can analyze large amounts of data. It can scan an entire model, detect issues, and highlight them in seconds. Manually, that could take hours, even days. This ability to summarize and surface insights so fast is a huge productivity boost, while still supporting the rigor MBSE requires.  

“An engineer designing a controller for a car or an aircraft wants to think about the solution, not spend mental energy clicking through menus. You can ask AI it to make this for you, and it handles the repetitive steps.”  

Walter van der Heiden, CEO Europe, SodiusWillert

3) What future use cases do you see where AI will have a significant impact on MBSE projects, and what are your predictions for how these might unfold? 

Andy: One major use case is model analysis. You can ask AI to examine a model and answer specific questions, like where potential issues lie, how much traceability is missing, or even just summarize its intent. AI excels at reading and understanding large volumes of information. With access not only to your model but also to broader knowledge bases, it can surface insights far more quickly than an engineer working alone.  

Walter: Another important use case is reverse engineering. Many teams have large amounts of source code that are poorly documented or no longer understood. Traditional reverse engineering in tools like Rhapsody often produced overwhelming results without adding clarity. With AI, it’s different. AI can read source code, understand its purpose, restructure it, and present it in the form of a clean, understandable model. What used to generate confusion now produces clarity.  

Could AI analyze code and deduce the requirements that guided its implementation?

Walter: Absolutely. AI can generate requirements directly from code or from existing models. When we first experimented with AI Modeling Assistant for IBM Rhapsody, I gave it a piece of old code labeled stopwatch.cpp. I asked the AI to reverse engineer it. Not only did it produce a clean model, but it also extracted the underlying requirements and documented them. That's something engineers have wanted for years: a way to recover both structure and intent from legacy code.  

Andy: Indeed, during our initial tests, one of the prompts was simply: “Build me a model of a stopwatch”. That was all. The AI not only built the model but also wrote the requirements and linked them to the model elements. It even added traceability automatically. We hadn’t asked for that, but it did it anyway.  And it did well! Another great benefit is documentation. Engineers rarely add thorough descriptions to models, even though they should. AI, however, generates meaningful descriptions as it builds models – or it can read your model and add missing descriptions for you. This improves clarity and communication within teams without adding extra work.  

4) Creating a system model from scratch has always been challenging and resource-heavy in engineering design.  How do you see AI changing that reality for systems and MBSE engineers? 

Andy: This is one of the most immediate and concrete use cases. Companies often start here when thinking about AI in MBSE. They have documentation, existing assets, and a set of requirements, and they don’t want to start from a blank page anymore. AI can process all of that and deliver a starter model, whether for a new system or a variant. It saves engineers from manually entering every single element while also interpreting the requirements.  So, AI can read requirements and generate a draft model. 

With AI support, even junior engineers can create models in SysML or UML, gaining guidance along the way without needing deep prior knowledge, right?  

Andy: Yes, indeed. For junior engineers, AI acts like a kind of mentor. It can explain complex concepts like proxy ports by building a model and pointing out mistakes in models they’ve built. It’s like having a dedicated teaching assistant that works whenever you do and has all the time in the world to help. And for senior engineers, it’s like having an army of junior engineers, eager and always ready to do the repetitive tasks, so they can focus on higher-level design decisions.  

Walter: But AI isn’t a human with gut feelings and proper common sense. It can propose logical solutions, but they still need to be reviewed. That’s why engineering judgments remain critical.  

That reinforces the point: AI does not replace human eyes, human mind, or human expertise.  

Walter: I like to compare it to early navigation systems from 25 years ago. They could tell you “Turn left here” or “turn right there”, but you still need basic knowledge of your route. If you typed in a destination, you had to know whether it was realistic. AI is similar. It can guide you and automate many steps, but you must monitor the outcome. Overlooking validation means forgetting what engineering is really about.  

“I might know exactly what I want to put in a model but doing it manually slows me down. AI can add content much faster and find what I need instantly.”  

Andy Lapping, Technical Fellow, SodiusWillert

5) In highly regulated industries like automotive, projects require end-to-end traceability. How can AI help identify gaps or inconsistencies that might otherwise put compliance at risk? 

Walter: One of AI’s most valuable contributions in regulated industries is strengthening traceability. It can automatically establish and verify links between requirements, model elements, and code. That makes it easier to spot where a connection is missing, something that might slip past even the most experienced engineers. In our stopwatch examples for early AI Modeling Assistant testing, we asked the AI to generate requirements and immediately add the traceability links. It did it perfectly. 

And is this directly relevant for compliance audits, like ASPICE or ISO 26262? 

Walter: Yes. AI can help prepare such audits, but I don’t think it will replace human-led assessments; auditors still expect people in the room. However, AI can be a powerful helper. Before an ASPICE assessment, for instance, you can provide the AI assistant with both your project information (models, requirements, code, etc.) and the compliance rules. It can then highlight areas where you’re solid and flag where you need improvement. That way, the formal audit process becomes more predictable and less pressuring.  

Andy: That’s essentially part of the model analysis use case we mentioned earlier. It’s not just assessments; it’s any kind of review cycle. Before any review, you could run the AI assistant to analyze the model against the specific criteria of that review. It can then answer a key question: “Am I on track, or not?”. And if not, it helps identify what needs to be fixed before the formal review.  

6) What would you say to engineers who still worry AI might introduce more risks than benefits? 

Andy: AI doesn’t inherently bring any more risk than not using AI.  The real risk is blindly trusting AI output without verification. AI can still put garbage into a model just like you can – it just does it faster. And AI never admits it doesn’t know something unless you ask it to cite its sources; it can infer or generate things that aren’t accurate. I even had one, once admitted (when challenged) that it "completely made that up". If no one checks the results, that’s when problems occur. The key is to review what AI delivers.  

Walter: The role of an engineer is shifting. You become more of a manager of output. You oversee, validate, and add your expertise where needed. The upside is that AI takes away the repetitive “click after click" tasks that consume so much time. Tools like IBM Rhapsody are very powerful, but working with them becomes much more efficient and pleasant when AI handles the heavy lifting.  

7) Looking to the few years (even months) ahead, where do you see the biggest opportunities for AI in systems engineering? 

Andy: The real opportunity is across the entire lifecycle, not just one part of it. Today, with AI Modeling Assistant for IBM Rhapsody, we’re focusing on IBM Rhapsody, but the same approach could extend to requirements, testing, hardware modeling, and other domains. Eventually, AI could connect all of these pieces into a holistic view. 

Walter: Exactly. And it won’t stop there. It could reach further into production and even asset management. I'm really excited to see what the future holds; everything is changing so fast! 

🤖 Walter and Andy summed it up well: AI won't stop at MBSE. The blank page effect in modeling may be ending, but the story of AI in engineering is only the beginning! 

To learn more about AI Modeling Assistant for IBM Rhapsody or discuss how AI can support your MBSE workflows, feel free to contact our team at SodiusWillert.