Natural Language Integration for CAD Automation

PTC is releasing the FeatureScript MCP Server within its Onshape CAD and PDM platform to enable engineers to create custom computer-aided design functions via natural language processing. This artificial intelligence integration targets industrial manufacturing and product development by automating recurring engineering workflows without requiring specialized programming knowledge.

Automating Design Workflows Through Large Language Models
The FeatureScript MCP Server utilizes the Model Context Protocol (MCP) to link coding-capable large language models (LLMs), such as Claude, ChatGPT, and Gemini, directly to the proprietary CAD automation language utilized by Onshape. Instead of manually writing code, users define the required design functionality via natural language inputs. The artificial intelligence system then drafts, tests, debugs, and refines the FeatureScript code until the custom tool meets the specified geometric or operational requirements.

This approach addresses the industrial demand for scalable engineering capabilities within a product lifecycle data ecosystem, allowing technical teams to transition from standard text-to-CAD generation to a text-to-code-to-CAD pipeline. David Katzman, EVP and General Manager of Onshape and Arena at PTC, noted that this architecture allows users to develop engineering capabilities specifically tailored to an enterprise's individual products and manufacturing processes.

Reusable Engineering Assets for the Product Data Ecosystem
Capturing and operationalizing engineering expertise is a critical component of managing complex design lifecycles. By compiling proven design methods into standalone, reusable CAD features, manufacturing enterprises can standardize workflows across distributed teams and projects. Because the generated code functions as a permanent FeatureScript tool, engineers can apply these automated solutions continuously without triggering recurring interactions or usage costs associated with ongoing LLM prompts.

Jeff Hojlo, Research Vice President of Industrial Ecosystems at IDC, indicated that adopting artificial intelligence for engineering shifts the focus toward capturing and operationally leveraging technical expertise rather than solely generating new geometric models. By building a unified product data foundation, enterprises can accelerate transformation processes, manage structural complexity, and maintain strict regulatory compliance standards across their supply chains.

Additional Context
This section details technical specifications and competitive benchmarking not included in the original news release.

Traditional CAD automation systems, such as Siemens NX Open or Dassault Systèmes SOLIDWORKS API, typically require specialized syntax knowledge in languages like C++, Python, or VBA to create custom design features. While artificial intelligence assistance can be applied to these legacy systems via external integrated development environments (IDEs) like GitHub Copilot, the Onshape FeatureScript MCP Server integrates the LLM protocol directly into the cloud-native CAD environment.

This direct protocol linkage addresses the latency and interoperability barriers frequently found in desktop-bound CAD architectures. By standardizing code execution directly on the server side rather than relying on local client installations, the MCP deployment establishes a unified, objectively measurable benchmark for enterprise-wide CAD tool deployment. This method eliminates the version-control friction typically associated with distributing custom automation scripts across decentralized engineering departments.

Edited by Aishwarya Mambet, Induportals Editor, with AI assistance.

www.ptc.com