Trusted AI Arrives in Embedded Systems Development
With the R2026a release of MATLAB and Simulink, MathWorks is weaving generative AI directly into the heart of embedded systems development. The headline additions—Simulink Copilot and Polyspace Copilot—are designed to accelerate design and verification without sacrificing rigor, traceability, or trust. Rather than acting as generic chatbots, these MathWorks AI tools are tightly grounded in users’ models, organizational processes, and official documentation. The goal is to help teams understand designs faster, catch issues earlier, and apply development and verification workflows more consistently across projects. This strategy reflects a broader shift: engineering leaders want AI that measurably improves productivity while upholding safety-critical standards. By embedding copilots into tools engineers already rely on, and by connecting MATLAB and Simulink into agentic workflows, MathWorks is positioning AI as an integral, auditable part of the modern engineering toolchain.
Simulink Copilot: AI Assistance for Model-Based Design
Simulink Copilot targets teams using Model-Based Design, where complex models can span thousands of blocks and subsystems. Grounded in the user’s existing Simulink models and team processes, it can generate natural-language explanations of model behavior, answer detailed questions, and help users locate relevant components quickly. This context-aware guidance is particularly valuable in embedded systems development, where engineers must trace requirements through intricate control logic and signal flows. Simulink Copilot also helps isolate problematic sections of a model, suggest remedies, and guide next steps, effectively acting as a design-side engineering assistant. Beyond troubleshooting, it can execute standardized tasks to enforce consistent modeling and verification practices across teams. For organizations under pressure to shorten development cycles while maintaining compliance, Simulink Copilot offers a way to scale best practices and reduce onboarding time for new engineers working on complex embedded architectures.
Polyspace Copilot and Polyspace as You Code: Smarter Static Analysis
On the software side, Polyspace Copilot focuses on embedded code verification, where static analysis findings can be numerous and difficult to interpret. By leveraging Polyspace analysis results, the copilot explains warnings, clarifies root causes, and suggests practical fixes, helping developers resolve issues more efficiently. This is especially important for safety- and mission-critical embedded systems, where understanding each defect’s impact is crucial. Complementing this, Polyspace as You Code brings continuous C and C++ checking directly into the coding workflow, including for code generated by AI-assisted tools. Developers can detect rule violations, defects, and vulnerabilities as they type, tightening the feedback loop between implementation and verification. Together, these capabilities shift defect discovery earlier in the lifecycle, reduce noise in static analysis reports, and support a more unified, disciplined approach to software quality across development, testing, and verification activities.
From Design to Production: Unified AI-Enhanced Workflows
MathWorks’ broader R2026a strategy is to make AI a continuous companion from early design through verification to production. In addition to MATLAB Copilot, Simulink Copilot, and Polyspace Copilot, the company is integrating MATLAB and Simulink functionality into agentic workflows via MATLAB MCP Core Server and MATLAB Agentic Toolkit. This allows organizations to orchestrate automated tasks—such as running analyses, generating reports, or triggering tests—within existing toolchains. Enhancements across the Polyspace family, including a new unified desktop application, custom checkers in Polyspace Bug Finder, and software-sanitizing functions in Polyspace Test, further streamline software quality management. For embedded systems development teams, the net effect is a more consistent and traceable pipeline: models and code can be designed, analyzed, and refined with AI assistance at every stage, while maintaining the discipline required for complex, safety-critical engineered systems.