New SolidWorks Add-in LAD Uses AI for Design Automation

Developers Will and Jorge have introduced LAD (Language-Aided Design), a specialized add-in for SolidWorks that leverages Large Language Models (LLMs) to automate the creation of CAD designs. The tool addresses a specific limitation in current engineering software: the inability to generate 3D modeling outputs directly from text prompts.

LAD functions as a bridge between conversational inputs and complex engineering tasks. It allows users to create sketches, features, assemblies, and macros simply by describing them. The system is designed to integrate deeply with the SolidWorks environment, providing the AI with necessary context through automatic screenshots and feature tree parsing. Despite the current limitations of LLMs in 3D object creation compared to code generation, the developers believe the technology will improve significantly in the coming months.

The development of LAD stems from the creators' backgrounds in software engineering, where tools like Claude Code and Cursor have become dominant. While exploring CAD systems, the developers identified a significant disparity: software development has embraced AI-driven text-to-code capabilities, but mechanical design lacks equivalent tools for text-to-3D modeling.

According to the developers, major CAD systems currently offer no native way to go from a text prompt to a modeling output. LAD aims to fill this void by acting as an intermediary layer. The developers acknowledge that in their testing, LLMs are currently not as proficient at generating 3D objects as they are at writing code. However, they express strong optimism that these capabilities will evolve rapidly in the upcoming months and years.

LAD operates as a comprehensive toolkit within the SolidWorks environment, translating natural language and uploaded documents into actionable CAD geometry. The add-in provides the LLM with a wide array of specific functions to manipulate the design environment effectively.

The system is built on several distinct tool categories designed to handle the full spectrum of mechanical design tasks:

• Part Creation: Dozens of tools allow the LLM to generate sketches, extrude features, and modify parts.
• Assembly Management: Specific tools enable the AI to organize individual parts into full assemblies.
• File Operations: The LLM can create, save, search, and read SolidWorks files and documentation directly.
• Macro Automation: Tools for writing and running macros are included, supported by an API documentation search feature.

To ensure the AI understands the current state of the design, LAD employs advanced context awareness mechanisms. The system automatically captures screenshots and parses the feature tree, feeding this visual and structural data back to the LLM. This allows the AI to make informed decisions based on the user's current progress rather than operating blindly.

Safety and user control are also central to the architecture. The developers implemented a checkpointing system that allows users to roll back unwanted edits, mitigating the risk of AI errors. Furthermore, a permissioning system determines which commands require user approval before execution, ensuring that critical changes are not made without oversight.

LAD is currently available for users to try, with the developers actively soliciting feedback to refine the tool's utility. The release represents an early step toward integrating generative AI into mechanical engineering workflows, a sector that has traditionally relied on manual parametric modeling.

The project highlights a growing trend of applying LLM capabilities to specialized, non-coding domains. As LLMs continue to advance, tools like LAD may fundamentally change how engineers interact with CAD software, shifting the focus from manual geometry creation to high-level design intent.