Concept for the development of Design Insight CAD using generative AI.

We developed a "Design Insight CAD System" that utilizes generative AI to enhance design rule validation in manufacturing. Traditional systems require manual analysis of design rule documents and complex coding for rule implementation, which are time-consuming and dependent on expert knowledge. To address this issue, the proposed system automates two key processes: extracting design rules from know-how documents and generating digitalized rules in Python using a standardized function database. We evaluated the impact of prompt design and input size on rule extraction accuracy, confirming that optimized prompts and reduced token counts significantly improve precision and recall. Additionally, we analyzed 55 manufacturing processes across five categories—removal, molding, joining, treatment, and automation—to identify common elements that support rule generalization and prioritization. Comparative testing showed that the use of generative AI reduces document analysis time from 5 minutes to 15 seconds and digitalized rule generation time from 2 hours to under 2 minutes. This study demonstrates that integrating generative AI into CAD systems can streamline design validation, reduce development time, and support frontloading in product design.