Artificial intelligent design technology for shear wall structures holds promise in enhancing design efficiency and addressing the tedious and repetitive nature of conventional design work. This technology has experienced rapid development in recent years. However, based on deep learning, the existing design methods for shear wall structures face challenges in simulating the entire process of real structural design. As such, improvement in terms of safety (mechanical performance) and cost-effectiveness (material consumption) becomes necessary. In response to these issues, this study proposes an intelligent design system for shear wall structures based on a large language model (LLM). The system utilizes the LLM as the core controller, to facilitate interaction with engineers and interpretation of their language descriptions into computer-executable code. The core controller then calls upon corresponding structural generation and optimization methods to improve the design outcomes. Additionally, during the process of structural optimization, the system integrates mechanical performance, material utilization, and empirical rules to simulate the engineer’s structural adjustment workflow, thereby improving the quality of design. The proposed system showcases its proficiency in successfully executing the comprehensive conversion process of architectural drawings, structural schemes, and analysis models, achieving an automated and highly efficient design approach for shear wall structures. Analysis and verification of multiple cases demonstrate that this system can improve design speed by more than 30 times (approximately 1 h as opposed to 30 h using traditional methods), while ensuring the safety and economic viability of the proposed design solutions. As a result, this study provides valuable insights for the realization of automated structural design.