High-Accuracy Text-to-CAD/CAM/CAE Generation Using Generative AI: The Archgen Framework

We introduce ArchGen, a state-of-the-art Text-to-CAD framework capable of generating complex, parametric 3D architectural and structural models directly from natural language prompts. Unlike conventional text-to-CAD systems that often struggle with design complete- ness and structural accuracy, ArchGen employs a multi-stage orchestration pipeline including prompt validation, CAD generation, execution checks, iterative refinement, and quality assess- ment. Leveraging retrieval-augmented generation with caching and cross-stage insight sharing, the system consistently produces highly detailed models, such as multi-story buildings, struc- tural reinforcements, and layout-precise assemblies. The outputs demonstrate architectural re- alism, supporting repetitive elements like floors, grids, and reinforcement details, making them directly usable for professional prototyping and design visualization. Experimental evaluation confirms that ArchGen achieves superior validity, geometric consistency, and requirement ful- fillment compared to baseline approaches. By bridging natural language understanding with intelligent design automation, ArchGen sets a new benchmark for CAD generation, accelerating workflows in architecture, structural engineering, and industrial design.