Sponsored Content
Richard Rovner & Udo Gohier, MathWorks
“Engineering and IT need to communicate effectively to successfully manage the shift”
In this interview, Richard Rovner and Udo Gohier explain how MathWorks supports automakers and suppliers in successfully navigating the cultural and technological transformation toward the Software-Defined Vehicle.
A key driver in comprehensive automotive development is the vision of a genuine Software-Defined Vehicle. How does MathWorks support automotive companies towards this target image?
Rich Rovner: The automotive industry is transforming into a software industry and that’s a major transformation. It’s difficult, it’s challenging, and it can be painful, but it also opens up enormous opportunities. MathWorks has been a strong presence in the automotive industry for decades, and we expect that to continue. We provide an engineering platform that supports this transformation end to end – from concept, into design and development, and on to production and operation. Our role is to deliver a platform that integrates into the entire workflow and helps companies accelerate their development while maintaining quality and safety. A key success factor in this transformation is effective collaboration between engineering and IT organizations. From a platform and service perspective, we see it as our mission to enable and facilitate that collaboration. Our tools and services are designed to make cross-functional teamwork more seamless and efficient.
Udo Gohier: The complexity that comes with realizing the Software-Defined Vehicle is enormous. As Rich mentioned, it’s not only a technological transformation — it’s also about mindset, systems engineering, and culture. Engineering and IT need to communicate effectively if companies want to successfully manage this shift. That means organizations must break down silos and enable collaboration between traditionally separate domains. We have seen some good examples where companies are working to design the next generation of organization to master this transformation and arrive at the digital product and digital company vision.
What is the reason why for many OEMs the road to a smoothly operating software factory is so bumpy?
Rovner: At a high level, this kind of change is inherently difficult and often painful. The automotive industry faces increasing complexity, growing competition, and higher demands for speed to market — all at once. To succeed, companies need a vertically integrated approach to technology that connects everything from hardware to software and systems engineering. But that’s easier said than done. OEMs must ensure safety and quality while transforming their organizations, and that means bringing together very different worlds: the engineering and engineered systems mindset and the IT mindset. These disciplines have traditionally worked in parallel, not in unison. Bridging them requires not only technical alignment but also cultural change.
Gohier: Exactly. And for traditional OEMs, there’s an additional layer of complexity: legacy environments. Many existing software processes were developed and optimized within isolated departments over decades. Now, these need to be transformed into an integrated, systemized environment. As Rich said, the real difficulty lies in the details. It’s easy to talk about “organizational transformation” in theory, but executing it in practice is extremely complex. The friction often arises between traditional departmental engineering structures and the ambition to create larger, more integrated software factory systems — which are frequently IT-led. The key is dialogue — and it needs to start early. If engineering and IT only begin to align late in the process, the result can be a new kind of silo or even a new “monolithic” system, rather than a truly integrated software organization. What we observe in successful transformations is that the necessary collaboration between these domains begins right from the start of developing modern software factory concepts.
How is the relationship between IT and engineering evolving in automotive development? And what technological and cultural trends are driving the convergence?
Gohier: We clearly see that IT and engineering are starting to look more closely at each other and recognize mutual benefits. Traditionally, there’s been a certain distance between the two worlds. Sometimes, IT departments tend to overlook the decades of engineering knowledge that have gone into building automotive systems. On the other hand, engineers have often regarded IT as a mere enabler, a supplier of infrastructure and digital capabilities. And that’s changing. We now see growing appreciation on both sides, with IT and engineering beginning to view each other as equal partners and knowledge stakeholders. If both approach collaboration with the mindset that “one plus one equals three,” the outcome can be much greater than the sum of its parts. A good example is the convergence of software development practices. Engineers often think in terms of embedded systems with tight constraints, while IT teams are used to cloud-based and enterprise architectures. Modern methodologies such as DevOps can help bridge these two worlds combining the rigor and validation discipline of engineering with the flexibility and scalability of IT systems. At MathWorks, we recognized this trend several years ago and have been working to integrate our software platforms — MATLAB, Simulink, and Polyspace — to cover the entire V-cycle of software development. This includes requirements management, modelling and simulation, functional and logical testing, deployment, and continuous improvement in operation through digital twins.
Rovner: The key to success lies in mutual understanding. IT organizations need to grasp the challenges engineers face — the platform complexity, safety and quality requirements, regulatory pressures — and appreciate the rigor these demands impose on development processes. At the same time, engineering teams should recognize the value IT brings: expertise in platform scalability, cloud infrastructure, cybersecurity, and enterprise integration. All of these are essential in the transition to Software-Defined Vehicles.
Balancing speed, quality and scalability is the real challenge
How can the ROI of engineering software be communicated effectively to IT and finance departments?
Gohier: I think at this point, no one in a leadership position needs to be convinced that software is essential to the future of the car. We’ve clearly passed the stage where companies still questioned whether there is a return on investment in software and services. However, what we do observe, especially in times of tighter budgets, is a much stronger focus on how to measure that return. Many OEMs and suppliers have made multi-billion-euro investments into software departments and digital architectures. Now, the question being asked more frequently is: What exactly constitutes a good ROI? That’s not always easy to define, particularly amid the ongoing disruption caused by the shift toward the Software-Defined Vehicle. But there are indicators that help make ROI tangible. For example: What is the degree of automation within your software factory? What is the quality of the code being produced? How effectively are models being reused across projects? How scalable is your overall software system? These metrics can provide valuable insight into both efficiency and quality gains. We already see customers using such indicators – though, in many cases, they could be applied even more rigorously.
Rovner: I completely agree. There have been enormous investments in technology, and companies naturally want to understand their return. The productivity improvements achieved through technologies like Model-Based Design and Software-Defined Vehicle architectures are significant. In the long run, these investments make not only engineering teams but entire organizations more productive. And it’s not just about internal efficiency, it’s about the end product. When you deliver better software, you deliver a better vehicle. That means happier, more loyal customers and ultimately higher revenue. Improved quality reduces defects, recalls, and warranty claims – all of which directly impact the bottom line.
At the same time, you can bring competitive products to market faster, enhancing your top line. So there are both top-line and bottom-line advantages to be gained from effective investment in technology and transformation.
Gohier: Exactly. And while we’ve discussed many of the challenges, there are also great success stories. In Germany, for example, companies like Bosch and BMW have made remarkable progress in systematizing software development. Looking to China, we see impressive cases at Geely and Zeekr, where they’ve built highly efficient software development ecosystems that already manifest in their products. These examples show how far the industry has come. People often focus on speed, moving from traditional 72-month development cycles to, ideally, six-week update cycles via over-the-air delivery. But speed alone isn’t everything. The real challenge is finding the right balance between speed, quality, and scalability. You can achieve excellent quality in a vertically integrated system but replicating and scaling that quality across an entire organization is another matter. The companies that manage to do both – accelerate development while maintaining quality and repeatability – are the ones truly mastering this transformation. And that’s where MathWorks contributes as well. Our continuous investment in end-to-end development platforms helps customers achieve exactly this balance: speed, execution quality, repeatability, and scalability.
Why is simulation a key lever for efficiency and quality in automotive development?
Rovner: Simulation has been an integral part of engineering for a long time and we continue to invest heavily in it, because it enables so much. With simulation, engineers can explore a much wider design space, perform virtual prototyping, validate against requirements, and test different scenarios long before any physical prototype exists. You don’t need actual hardware or chips to start testing. The ultimate vision is to simulate the entire vehicle. But even today, you can simulate at every level: from individual components to systems and systems of systems. Simulation connects the entire workflow, from requirements through verification and validation, to deployment, and allows you to iterate much faster. Another key advantage is scalability. You can simulate on a local workstation or scale up to the cloud, depending on your needs. It gives engineers the flexibility to perform targeted studies or large-scale system evaluations – all within a consistent environment. Simulation also extends beyond development into the operational phase of the vehicle through the concept of the digital twin. Once a vehicle is on the road, operational data can be fed back into the simulation environment. Engineers can compare real-world performance with the digital model, update it accordingly, and even deploy improved models to vehicles in near real time.
In this context, how do you balance “China speed” with “automotive quality”?
Gohier: First, remember, it’s not just quality but also safety. The faster you move, the greater the risk of compromising quality and safety. That’s the core challenge. You simply cannot sacrifice safety in pursuit of speed. Engineering organizations are pushing the envelope here, but it’s especially difficult for established manufacturers with decades of legacy systems and processes. Greenfield operations, by contrast, can design ideal structures from scratch – the perfect organization, process, and workflow – and therefore ramp up speed more easily. With the increasing complexity of software, maintaining a high standard becomes harder, yet it remains non-negotiable. What we often observe is that companies struggle to balance speed with safety, regulation, and homologation requirements. These are natural constraints that can slow down development. That said, we’ve seen remarkable progress — particularly in Europe and Germany — where production cycles have shortened dramatically. From my observation, development timelines have shrunk by roughly one-third compared to six or seven years ago. That’s a tremendous achievement.
Rovner: I completely agree and this is exactly where Model-Based Design, modelling, and simulation play a crucial role. They provide a way to achieve both speed and quality without compromising safety. With a modelling and simulation approach, you can automate testing, design and run virtual scenarios, and synthesize data to allow you to test for low-probability scenarios and edge cases that might be impossible in the physical world. You can validate designs against requirements, ensure compliance with regulations, and verify safety and reliability all within a virtual environment, long before vehicles hit the road.
How do you enable integration of your engineering tools into CI/CD pipelines?
Rovner: We’ve long focused on delivering software that’s open, interoperable, and easily integrated into existing toolchains, including modern cloud platforms and CI/CD environments. Our goal is to make sure that MATLAB, Simulink, and their outputs fit seamlessly into enterprise IT infrastructures and cloud-native workflows. A good example is automated testing and code generation. With our tools, engineers can automatically generate production-quality code directly from models and feed it into CI/CD systems for continuous testing and validation. This enables a smooth flow from model development to deployment, ensuring consistency, quality, and traceability across the entire process. We continue to invest heavily in making our products interoperable with major CI/CD systems and cloud environments. That includes robust API connectivity and tight integration with hyperscaler platforms, so that our users can run their workflows seamlessly in the cloud or on enterprise infrastructure.
Gohier: In today’s complex toolchain landscape, no single vendor can cover everything from A to Z. That’s why we deliberately design our software to be open and to integrate with other systems and standards that are already established in the market. If we didn’t do that, if we built a closed ecosystem, we would effectively isolate ourselves and make it harder for our customers to implement efficient, end-to-end workflows. Instead, we embrace a platform thinking approach, ensuring our tools work as part of a larger ecosystem where interoperability drives both flexibility and scalability.
What is the MathWorks stance on open source and how do you interact with corresponding components and the community?
Rovner: We actively support open-source technologies and communities, and we design our products to interoperate seamlessly with open-source environments. For example, users can import models from open-source platforms into MATLAB and Simulink, and they can also export code from our tools so that it runs in open-source ecosystems outside of the MathWorks environment. We’ve long provided open interfaces and integrations for widely used standards and middleware such as Adaptive AUTOSAR or Android Automotive ensuring that our tools work smoothly within these frameworks. Our goal is to give engineers flexibility: you can use MATLAB and Simulink as your core development foundation and connect open-source components around them, or integrate our tools as one part of a broader open-source toolchain. We also collaborate directly with the open-source community to maintain interoperability and ensure our technologies fit into modern, mixed ecosystems. Ultimately, our philosophy is openness, to provide customers with the freedom to combine proprietary and open-source technologies in the way that best supports their innovation.
How is Generative AI transforming the engineering design process?
Rovner: It’s an incredibly exciting time. We see traditional AI, generative AI, and agentic AI all playing a role in enhancing the work of engineers. What these forms of AI allow is a shift to a higher level of abstraction. They can automate repetitive or time-consuming parts of the workflow, enabling engineers to explore a much wider design space without getting bogged down in low-level coding or implementation details. The potential here is huge: AI can assist from the earliest design stages all the way through to code generation and execution on embedded devices.
Where are the limitations?
Rovner: It’s essential to be realistic. These technologies are not perfect, they make mistakes. And while it’s one thing to get an incorrect answer in a web search, it’s something entirely different when a generative model produces flawed code that ends up in an embedded system. That’s why Model-Based Design is such a critical framework: it allows engineers to verify, test, and validate what AI produces, both automatically and manually, before it ever reaches production. We like to say: AI can automate workflows and generate code, but Model-Based Design proves that it works. That combination, as we call it AI-assisted Model-Based Design, represents, in our view, the future of engineering.
Gohier: Exactly. The future lies in combining automation with the physical rigor that engineers rely on and trust. Particularly in areas like physical modelling, there’s still a lot of work to do and that’s where AI-assisted Model-Based Design shows its full potential. When engineers release models downstream into production workflows, they need absolute confidence in their accuracy and safety. AI-assisted Model-Based Design helps achieve that by freeing engineers from routine or administrative tasks so they can focus on the areas that truly require their expertise.
AI-assisted model-based design is the future of engineering
How is AI changing the skillset required for engineers and what new roles or competencies are emerging?
Rovner: There’s no question that AI is reshaping the skillset engineers need and it’s happening right now, not sometime in the future. With AI becoming part of the engineering workflow, new competencies are required to use it effectively, responsibly, and safely. First and foremost, engineers need what I’d call AI literacy. They have to understand how AI enhances their workflows, but also recognize its risks, limitations, and potential failure modes. Applying the rigor of established engineering practices, like testing, verification, or validation, remains critical to identifying, minimizing, and ultimately eliminating errors before they propagate through the system. Second, engineers need to think in systems. AI integration requires a systems mindset — understanding how different components interact, how data flows, and how to design architectures that keep safety, reliability, and compliance at the core. Third, agility is becoming an essential skill. The pace of change in AI is extraordinary. Engineers must be able to quickly evaluate and adopt new technologies — but always in a way that aligns with the principles of reliability, traceability, and provability that define good engineering. And finally, collaboration is more important than ever. We’ll see AI specialists joining traditional engineering teams, and closer partnerships with IT, data science, and software development functions. Success will depend on how effectively these diverse disciplines work together.
How is MathWorks embedding GenAI into its products?
Rovner: We’re taking a very focused and deliberate approach. This year we introduced MATLAB Copilot, an AI assistant designed to help users understand and write MATLAB code more efficiently. Next year, we plan to extend this concept with Simulink Copilot and Polyspace Copilot. These tools are being developed methodically and responsibly. The goal is to help users understand, interpret, and ultimately generate code and models through natural interaction with AI assistants. Each step is carefully designed to ensure reliability and accuracy – qualities that engineers expect from MathWorks tools. Importantly, we’re building these capabilities with flexibility in mind. Different organizations and user groups – from automotive engineers to academic researchers – have varying levels of readiness for adopting AI. Our approach allows them to adopt at their own pace and under their own governance, maintaining control over how AI is integrated into their workflows. We’re also seeing a major shift in academia. A year ago, many universities were still skeptical about generative AI, but that resistance is fading quickly. Educators are now embracing it as a valuable tool for teaching and research, and we’re actively working with them to shape how AI can best support the next generation of engineers.
Gohier: For me, emphasis on responsibility cannot be overstated. When it comes to generative or agentic AI, we don’t simply follow trends, we take the time to model these innovations properly and ensure they meet our quality standards before release. Especially in areas like simulation and model creation, precision is everything. Engineers need to trust that our tools not only help them build and test models, but also detect and correct potential errors early in the process. That’s why we involve our customers early on, collecting feedback, validating use cases, and refining functionality in close collaboration. Only when the quality matches our standards do we release new AI-enabled features.
Let’s look ahead: As automotive development becomes increasingly software- and AI-driven as well as cloud-integrated, what do you see as the biggest opportunities and risks for OEMs and suppliers in the coming years?
Rovner: The biggest opportunity is undoubtedly the chance to build the software factory of the future. This is a transformative moment: the automotive industry is, in many ways, evolving into a software industry. That shift opens enormous potential for innovation, efficiency, and new business models. But, of course, it also comes with risks. The growing complexity of vehicles, intensified market competition, and the challenge of maintaining safety, quality, and reliability all increase the pressure on organizations. The key lies in approaching this transformation in a methodical, intelligent, and responsible way. At MathWorks, we believe that AI-assisted Model-Based Design provides the framework for managing these risks while realizing the full opportunity. It helps organizations systematize their development processes, maintain rigorous verification and validation, and balance the competing demands of speed, safety, and scalability.
Gohier: Building on that, I think the biggest opportunity – and simultaneously the greatest challenge – is to stay systematic. Moving from traditional departmental engineering to a truly systemized, model-driven approach is essential for long-term success. It’s tempting to try to do everything at once, but that’s rarely effective. Companies need to stay focused, choose their priorities wisely, and advance step by step. Identify where to invest your efforts, which systems, processes, or domains to transform first, and then follow through methodically before moving on to the next area. And just as importantly, ask for help and encourage collaboration between domains. No single department or discipline can master this transformation alone. The organizations that foster cross-functional cooperation between IT, engineering, software, and data will be the ones that succeed in building the next-generation automotive ecosystem.
This article is sponsored by:
Find more information on SDV development here.