Generative AI delivers the most value when it is deeply integrated with modeling and simulation

Seth DeLand, Product Manager, MATLAB Copilot

How is Generative AI fundamentally changing the way engineers approach early-stage design, particularly in exploring vast design spaces that were previously impractical to evaluate manually?

Generative AI is changing early-stage engineering design by accelerating exploration rather than replacing engineering judgment. At MathWorks, this shift is grounded in a simulation-first approach, where engineers use models of physical systems to explore alternatives safely and systematically before committing to hardware.

Generative AI capabilities such as MATLAB Copilot help engineers navigate and assemble complex modeling, simulation, and optimisation workflows that already exist within MATLAB and Simulink. Instead of manually stitching together tools or writing boilerplate code, engineers can more quickly explore control strategies, parameter variations, and design trade-offs inside a physics-based environment. By the way, the new generative AI copilots from MathWorks will be available soon.

This makes it practical to explore design spaces that were previously too time-consuming to evaluate manually, while still anchoring every outcome in deterministic models that engineers can simulate, verify, and validate before deployment.

How is Generative AI transforming engineering software development – especially for control systems, embedded systems, and industrial automation – and what risks or validation challenges remain?

In control systems, embedded systems, and industrial automation, Generative AI delivers the most value when it is deeply integrated with modeling, simulation, and code generation workflows. Within MATLAB and Simulink, Generative AI can assist engineers by suggesting control strategies, generating draft controller implementations, and explaining complex models or existing code.

A critical differentiator in our approach is that Generative AI is grounded in MathWorks documentation and established workflows. This grounding is essential for developing reliable and verifiable engineering software, because engineers must be able to understand where recommendations come from, trace them back to documented capabilities, and validate them through simulation and testing. By relying on documented APIs, workflows, and engineering methods, Generative AI supports transparency and traceability rather than opaque or unverifiable outputs.

The core validation challenge remains unchanged: AI-assisted results must be verified using simulation, testing, and established certification processes before deployment to real hardware. Generative AI accelerates development and understanding, but it does not replace engineering rigor – particularly in safety‑critical and industrial applications.

Design optimisation for lightweighting is a major advantage with reduced wastage. How meaningful are these gains in real industrial settings for achieving sustainability and net-zero goals?

Simulation-driven optimisation has already shown meaningful impact in real industrial settings, particularly when combined with AI-based optimisation techniques. By relying on digital models rather than repeated physical prototypes, engineering teams can reduce material usage, minimise waste, and explore more efficient designs earlier in the development cycle.

In industrial automation and machinery applications, simulation-based approaches allow teams to evaluate efficiency, emissions, and performance trade-offs before deployment. This makes sustainability considerations part of the engineering process from the outset, rather than an afterthought.

While sustainability outcomes ultimately depend on how organisations apply these tools, simulation-first engineering makes it practical to evaluate lightweighting, efficiency, and emissions trade-offs when design changes are least costly and most effective.

How much has Generative AI reduced design-to-prototype timelines, and what organisational or workflow changes are needed to fully realise these gains?

Generative AI contributes to shorter design-to-prototype cycles by reducing friction in engineering workflows rather than eliminating engineering steps. By helping engineers assemble workflows, generate draft code, and understand complex models more quickly, Generative AI shortens iteration cycles during early design and development.

To fully realise these gains, organisations need simulation-first workflows, reusable models, and well-established validation practices. Generative AI delivers the greatest value when layered on top of mature modeling, simulation, and automated code generation processes.

In practice, teams that already rely on digital twins and model-based design are best positioned to translate Generative AI productivity gains into faster development timelines.

As engineers increasingly rely on AI-generated designs and recommendations, how should organisations approach validation, certification, and accountability – especially in safety-c

In safety-critical industries, accountability cannot be delegated to AI. Generative AI at MathWorks is designed to operate within deterministic, validated engineering environments where every recommendation can be simulated, tested, and reviewed.

Engineering teams must continue to rely on simulation, formal verification, testing, and established certification processes to validate AI-assisted designs. Generative AI helps engineers explore options and work more efficiently, but responsibility for validation and certification remains firmly with the organisation.

The role of Generative AI is to increase engineering productivity and insight – not to bypass the rigor required to deliver safe, reliable, and certifiable systems.

(The new generative AI copilots from MathWorks will be available soon)

Seth DeLand is Product Manager for MATLAB Copilot.  He has also held roles at MathWorks as product marketing manager for machine learning and numerical optimization products and in technical support. Prior to joining MathWorks, Seth earned his M.S. in mechanical engineering from Michigan Technological University. His research was in mathematical modeling techniques for selective catalytic reduction on diesel vehicles to reduce NOx emissions.