MathWorks Adds AI Copilots to MATLAB and Simulink 

That latest update to MathWorks’ MATLAB and Simulink software includes new AI capabilities for embedded systems development. Release 2026a of MATLAB and Simulink includes AI-based copilot features.

The new Simulink Copilot supports model-based design, while Polyspace Copilot improves embedded software code analysis, which the company says can help engineering teams improve productivity while “maintaining rigor, traceability, and repeatability in their designs.” The new release also includes additional updates that help engineers design faster, fix issues earlier, and move more efficiently from development, through verification, to production. 

According to MathWorks’ Senior Principal Technologist Jason Ghidella, the need for these types of AI features is driven by the increased engineering complexity, which can make it difficult for new team members or stakeholders to get up to speed on these complex models. “Simulink Copilot helps you undnerstnd and work with these models quickly,” he says. “As you open the model it can give you direct advice about the structure, or what block to go to if you are lookign for a particular subsystem.”

The Polyspace Copilot also helps with onboarding, while also improving productivity without compromising the required rigor of embedded code analysis. 

According to a company press release, MathWorks is embedding copilots directly into the environments engineering teams are already using, such as MATLAB Copilot, Simulink Copilot, and Polyspace Copilot. At the same time, MathWorks has integrated MATLAB and Simulink functionality into agentic workflows through MATLAB MCP Core Server and MATLAB Agentic Toolkit. The company says this allows teams to understand designs faster, address software issues earlier, and apply development and verification workflows more consistently.  

“Engineering teams now have access to capabilities enabled by generative AI, and leaders need confidence that these translate into tangible engineering and business benefits,” said Avinash Nehemiah, Head of Product Management and Marketing, Design Automation at MathWorks. “In engineering design and software verification, productivity improvements cannot come at the expense of rigor, traceability, or trust. MathWorks is committed to delivering grounded AI tools for engineering that help teams move faster while preserving the discipline and confidence required to develop complex engineered systems.”

MATLAB Agentic Toolkit equips AI coding agents with expert knowledge and a live MATLAB connection to write idiomatic code, test and debug it, build apps, and use the full range of MATLAB toolboxes across evolving agent platforms. Image courtesy of MathWorks.

Simulink Copilot provides guidance tailored to the engineering context teams already use, the company says. The product can generate model explanations, answer questions about model behavior, and help users locate relevant blocks and subsystems in their models. By isolating issues, suggesting remedies, and guiding next steps, the copilot helps engineers move design work forward more quickly. Engineering teams can also use it to execute standardized tasks that support more consistent development and verification practices.  

R2026a introduces Polyspace Copilot and Polyspace as You Code. Polyspace Copilot provides guidance based on Polyspace analysis results to help engineers interpret static analysis findings, understand issues, and resolve them more efficiently. Polyspace as You Code enables developers to check C and C++ coding rules and identify coding defects and vulnerabilities as code is written, including code produced with AI-assisted tools. Together, these offerings help teams identify issues earlier and strengthen software quality throughout the development lifecycle.

“The big difference between Simulink Copilot and generic AI tools our customers might be familiar with is that we are reading the model structure and using MathWorks documentation for tha guidance,” Ghidella said. “It also opens up within the model itself. There is no context switching, which minimizes errors. With the Polyspace Copilot, the solution  is grounding its responses based on the results of the Polyspace formal analysis engine. The copilot can identify a defect and then explain why the problem has been identified as a defect, and then helps you resolve the issue.”

R2026a also includes three enhancements across the Polyspace product family: a new Polyspace desktop application for unified configuration and results management, extensions to Polyspace Bug Finder with custom checkers and coding standards, and software-sanitizing capabilities in Polyspace Test for dynamic analysis of runtime errors. Together, these enhancements provide a more unified workflow for software quality activities across development, testing, and verification.

R2026a’s updates across the MATLAB and Simulink product families include:

• MATLAB Course Designer: A new product that helps educators develop courses, courseware, labs, and assessments using MATLAB and Simulink.

• Simulink FMU Builder: A new product that creates standalone Functional Mockup Units from Simulink models and C or C++ code to support model exchange and integration workflows.

• MATLAB: Build and share interactive webpages with visualizations without installing MATLAB. Engineers can also manage Python environments and improve data exchange between MATLAB and Python workflows.

• Simulink: Access commonly used actions more efficiently through simplified, task-focused context menus. Users can also simulate C and C++ code within models without language limitations or additional wrappers.

• Wireless Network Toolbox: Model, simulate, analyze, and visualize wireless communication networks to evaluate end-to-end system behavior.

• MATLAB Test: Generate starter tests, equivalence tests, and tests from command history using MATLAB Copilot. Engineers can also run tests related to the current file to reduce unnecessary execution.

• Mapping Toolbox: Enhance geospatial analysis with 3D building visualization, image overlays, and raster map capabilities.

• Signal Processing Toolbox: Design and analyze digital filters with the new Filter Designer and Filter Analyzer apps, label time-frequency data, and extract signal features using enhanced interactive tools.

You can learn more about the new update here.

Sources: Press materials received from the company and additional information gleaned from the company’s website.