A new generation of engineering workflows aims to move beyond AI-assisted coding by enabling AI agents to build models, execute simulations, analyse results, and refine designs within validated development environments.
While most technology companies are racing to add AI assistants to their software, MathWorks is pursuing a different challenge: making AI trustworthy enough for engineering environments where mistakes can have real-world consequences. The company is expanding MATLAB and Simulink beyond traditional AI assistance by enabling AI agents to directly interact with engineering workflows, execute simulations, test designs, and iteratively improve results inside controlled development environments.
The problem MathWorks is trying to solve is one that many engineering organisations now face. Generative AI can rapidly create code, models, and design suggestions, but engineering teams still struggle with verification, traceability, compliance, and validation—especially in industries such as automotive, aerospace, semiconductor design, and industrial automation. In this exclusive conversation, Prashant Rao, Head of Application Engineering, MathWorks India, explains, “Instead of focusing solely on faster code generation, MathWorks is embedding AI into a framework designed to maintain engineering rigour”. The company’s latest efforts combine AI-driven automation with verification and validation workflows that engineers already rely on for safety-critical systems.
One of the most notable developments is MathWorks’ open-source agentic AI toolkit, which enables AI agents to interact directly with MATLAB and Simulink. This enables AI systems not only to generate engineering models but also to run simulations, evaluate outcomes, troubleshoot issues, and refine designs in a closed-loop workflow. A particularly intriguing capability demonstrated during the discussion was the ability to generate engineering system architectures directly from requirements documents. In one example, AI could interpret requirements and automatically create system-level models, dramatically reducing early-stage design effort.
The company is also making these capabilities available through open-source toolkits hosted on GitHub, with updates being released far more rapidly than traditional software release cycles. New features are being added every few weeks, signalling a shift toward a more agile AI innovation model.Beyond productivity gains, MathWorks believes the bigger opportunity lies in reducing one of engineering’s most persistent bottlenecks: the lengthy cycle of testing, verification, and validation that often stretches development timelines. The company sees AI helping automate large portions of this process while keeping engineers firmly in control of final decisions.
Another area attracting attention is cost efficiency. As enterprises increasingly experiment with large language models, concerns are emerging around AI operating costs and token consumption. MathWorks is developing workflow-specific AI skills designed to reduce unnecessary AI interactions while keeping engineering tasks grounded in domain-specific processes. The long-term vision extends beyond AI copilots acting as assistants. MathWorks is building an ecosystem where AI agents can become active participants in engineering workflows—designing, testing, simulating, and refining systems while operating within established engineering constraints.
As industries search for ways to balance AI-driven speed with engineering reliability, MathWorks appears to be betting that the future will not belong to AI that simply generates answers, but to AI that can prove those answers work.
