IEN Europe: Since not all readers are equally familiar with this topic, could you please briefly introduce Onshape and explain how it differs from a traditional CAD structure?
D. Henry: Onshape is a cloud-native CAD and product data management platform that runs in a browser or on a mobile device. There is no software to install, update, or maintain locally, which virtually eliminates the IT overhead associated with traditional CAD deployments. Conventional CAD relies on files, often with a separate system to manage revisions and access. Onshape was built around a central database. Engineers work on the same current design, with version history, permissions, and collaboration built in. Multiple people can work concurrently without passing files back and forth.
Another distinction is FeatureScript, the programming language behind Onshape’s modeling features. We make it available to users and the broader community so engineers can create their own parametric tools. AI makes it much easier to tailor those tools to a company’s products and processes.
IEN Europe: Artificial intelligence is being integrated into a wide variety of software areas. What are the advantages and disadvantages of the current options for engineers in product development using CAD?
D. Henry: AI is showing up in CAD in several ways. AI advisors can answer questions and help people learn a system. Assistants or agents can perform tasks using the software's tools and APIs. Discrete capabilities, such as Onshape's AI Quick Render, can accelerate a specific part of the workflow. I think we will see more of these in every CAD product. They save time, though each addresses a particular task.
There is a lot of interest in generating geometry directly from a prompt. The challenge is that a plausible shape is not necessarily an engineering-ready design. If the geometry is not constructed with the relationships, parameters, and design intent that a human engineer would use, it can be difficult to change, validate, or reuse. Engineers need to make controlled changes as requirements evolve.
What excites me most is text-to-code-to-CAD: using AI to help engineers create customized, deterministic design tools. The engineer contributes knowledge and evaluates the results; the code applies those rules consistently within the CAD system. That can change how a whole team develops products.
IEN Europe: Could you please explain how the FeatureScript MCP Server differs from this and what changes this might bring to an engineer's workflows?
D. Henry: The FeatureScript MCP Server connects an AI coding assistant to the tools it needs to write, test, and refine FeatureScript for Onshape. An engineer can describe what a design tool should do, inspect the result in CAD, identify what needs changing, and iterate with the assistant. The output is not a one-off shape produced by a prompt. It is a native custom feature that other engineers can apply, edit, and reuse.
This could be a specialized thread or fillet, but it can go much further. It can generate fully defined component geometry or coordinated systems of parts, apply engineering calculations and produce information such as cut lists. FeatureScript also gives the feature a native user interface: inputs can be presented in a dialog, checked for valid values, and refined so colleagues can use the tool without knowing its underlying code. If an input is invalid or a model fails, the feature can tell the engineer what went wrong, right where they are working.
Instead of modeling the same type of design over and over, an engineer can build a tool for that whole class of designs. The engineer still has to test the feature and decide whether it does the job. Once the feature works as intended, the expertise behind it can be applied again and again, with consistent results and an editable model.
IEN Europe: What advantages do the new capabilities and workflows offer when considering the entire company rather than just the individual engineer?
D. Henry: Companies have long captured their intellectual property in product designs: the models, drawings, and specifications. Now they can also capture it in the tools used to create and evaluate those designs. A custom feature can encode preferred construction methods, product standards, and manufacturing constraints, then make that knowledge available to the whole team.
And this goes beyond geometry. A company could build tools that calculate fluid volume, locate a center of mass, or assess acoustic reflections in a proposed space. It could combine those checks with configurable parts or systems of parts, guiding engineers as they make design decisions. A specialist can develop and validate the method; colleagues can then use it through an approachable interface. That helps newer employees contribute sooner and gives experienced engineers a common, repeatable starting point.
The payoff is that the next project does not have to start from scratch, or depend on finding the one person who knows how something was done. Teams can share and improve a library of tools as products and processes evolve. Each improvement becomes something the whole team can build on. Engineers spend less time repeating routine work and more time on problems that need their judgment.
IEN Europe: So, on the one hand, the text-to-code-to-CAD approach increases flexibility, lowers barriers to entry, and helps make existing knowledge systematically accessible. What might that look like in a concrete example?
D. Henry: Consider a manufacturer of garden accessories. Its engineers need hose threads across many injection-molded products and 3Dprinted prototypes, with different lead-ins, lead-outs, and manufacturing requirements. If the standard CAD tools do not cover those variations, an engineer can use the FeatureScript MCP Server to develop a custom thread feature and share it with the team. Everyone can then apply the same rules while keeping the geometry editable.
A manufacturer of wire shelves, racks, and baskets could go further. Its engineers could create a tool that takes a selected shape and dimensions, applies the company’s spacing and bending rules, and generates the wire layout, cut lists, and bend tables. That tool could serve an entire product line.
These are two fairly simple examples. The broader opportunity is to tailor CAD to the products a company makes and the processes it uses, without waiting for a software vendor to build every specialized tool. AI puts that kind of customization within reach of far more manufacturers. It is almost like having a CAD development team working on your specific engineering needs, with your own engineers defining and validating what the tools must do.
