In the rapidly accelerating landscape of modern product development, the traditional "document-based" approach to systems engineering is hitting a wall. As engineered systems grow in complexity—from software-defined vehicles to intricate, multi-domain industrial ecosystems—the reliance on static, text-heavy documentation is increasingly viewed as a liability.
In episode 687 of Amelia’s Weekly Fish Fry, host Amelia Dalton sat down with Becky Petteys of MathWorks to dissect this critical industry transition. The conversation explored the evolution of Model-Based Systems Engineering (MBSE), the integration of AI into design workflows, and why the future of engineering relies on moving from "alchemy" to a hard, evidence-based science.
The Core Problem: The Failure of Document-Based Engineering
For decades, systems engineering relied on a manual, document-centric paradigm. Engineers would draft specifications, manage requirements in spreadsheets, and rely on the accumulated experience of senior staff to predict system performance.
According to Petteys, this model is no longer tenable. "In the past, when engineered systems were much simpler, an experienced system engineer could look at a document and understand how a system would perform," Petteys explains. "Today, systems are too complex. There is simply too much for one person to absorb, understand, and predict."
The limitations of this approach manifest in several ways, most notably in "Ctrl-F change management"—a risky practice where engineers manually search for and update occurrences of a requirement within documents. This method is highly prone to human error and often fails to capture the ripple effects of a change across interconnected system domains.
Chronology of an Industry Pivot
The shift toward MBSE has been a gradual but inevitable trajectory, mirroring the industry’s adoption of Model-Based Design (MBD) two decades ago.
- The MBD Precedent (2000s): MathWorks initially pushed for the use of Simulink to model and generate code for control systems. At the time, the industry was skeptical, favoring manual text-based specifications. Today, MBD is the standard for safety-critical applications.
- The Complexity Explosion (2010s): As connectivity and software integration became central to product design, the "silo" approach to engineering began to collapse. Systems engineering was forced to move beyond documents to keep pace with rapid iteration.
- The Advent of SysML 2 (2025): The introduction of the SysML 2 standard has provided a technical foundation for centralizing system models. This allows for interoperability between tools, enabling engineers to share models across analysis and verification environments rather than keeping data trapped in isolated documents.
Supporting Data: Digital Twins and Performance Metrics
A central theme of the discussion was the strategic use of "high-fidelity digital twins"—not as historical records of existing machines, but as predictive representations of potential systems. By simulating these models early in the development cycle, engineers can identify unintended behaviors and evaluate trade-offs before physical prototypes are built.
Key Performance Impacts
- Decision Velocity: Organizations leveraging simulation and analysis have reported reducing the time required for early design decisions by up to 90%, compressing cycles from months to days.
- Defect Mitigation: By identifying requirement conflicts during the design phase, teams avoid the "worst-case scenario": discovering fatal flaws during late-stage integration.
- The "Chemistry" of Engineering: Using the analogy of the evolution of alchemy into chemistry, Petteys emphasizes the need for measurement. By quantifying complexity—measuring how structural and functional complexity impacts cost and schedule—engineering is transitioning from a "magical art" to a rigorous, data-driven discipline.
Official Perspective: MathWorks’ Vision for AI and Data Science
MathWorks views the intersection of MBSE, AI, and data science as the "force multiplier" for the next generation of engineers.
The Role of AI
Petteys distinguishes between "hype" and "utility" in AI. She highlights three practical applications:
- Reduced-Order Modeling: Using machine learning to capture the key dynamics of high-fidelity models, making them lightweight enough for system-level simulations.
- Automating Administrative Burden: Leveraging Agentic AI and Large Language Models (LLMs) to handle repetitive tasks, such as updating specifications, cross-referencing requirements, and extracting data from disparate sources.
- Impact Analysis: Utilizing GenAI to scan the "digital thread" of a project, allowing engineers to instantly understand the downstream effects of a requirement change.
Data Science as the Foundation
"It’s all data," Petteys remarks. Modern engineering requires massive datasets for model creation, validation, and analysis. Data science enables teams to visualize these complex relationships, mitigate risk through predictive analytics, and optimize systems for performance rather than relying on intuition.
Implications for Industry: The Software-Defined Vehicle
Perhaps the most vivid example of this evolution is the automotive sector. Automakers are moving away from hundreds of localized Electronic Control Units (ECUs) toward centralized, software-defined architectures.
This shift is breaking down the traditional wall between "systems engineers" and "software architects." When both groups share a common toolchain and a "single source of truth," the speed of innovation increases exponentially. Petteys cites the example of Bosch in China, where teams integrated systems and software development into one ecosystem. By performing virtual testing early, they were able to maintain system integrity even while specifications were in constant flux.
Overcoming the "Muscle Memory" Barrier
The transition to MBSE is not merely a technical challenge; it is a cultural one. "There is a lot of muscle memory involved," Petteys notes. "There is a lot of perceived risk."
For organizations looking to adopt these methods, the primary hurdle is overcoming the fear of the unknown. However, the cost of inaction is rising. In an era where competitors are using AI-assisted, model-based workflows to shorten time-to-market, sticking to legacy document-based methods is increasingly seen as the riskier strategy.
Recommendations for Engineering Leaders:
- Embrace Interoperability: Move away from the idea that all data must live in one "magic" tool. Instead, focus on knowing where the authoritative source of truth resides and using standards like SysML 2 to connect tools.
- Prioritize "In-the-Loop" AI: Ensure that AI is used to remove "busy work" while keeping human subject matter experts in the loop for critical decision-making.
- Quantify Everything: Adopt the "chemistry" mindset. Start measuring complexity and track the correlation between model-based validation and reduced late-stage rework.
Conclusion
The dialogue between Amelia Dalton and Becky Petteys serves as a roadmap for the future of engineering. The industry is clearly trending toward a more integrated, simulated, and data-centric reality.
For the systems engineer of 2026 and beyond, the path forward is clear: the era of the static, document-driven project is coming to an end. By integrating high-fidelity digital twins, harnessing the power of AI to automate administrative overhead, and treating engineering as a quantifiable science, organizations can achieve a level of efficiency and innovation that was impossible just a decade ago.
As Petteys concludes, the benefit is not just in the bottom line or the product’s performance—it is in the morale of the engineering teams, who are finally being freed from the drudgery of outdated, manual processes to focus on what they do best: solving complex problems.
