Research on Agentic AI-Driven Automation and Collaboration in Full Architectural Design Workflow for Personalized Spatial Generation Based on User Behavior Perception

Abstract

Architectural design has long relied on experience, and there is often a gap between the design intent and the actual implementation of the project. This study focuses on two main lines of autonomous intelligent AI intervention in architectural design, namely full-process collaboration and personalized space generation, and examines the possible paths for reconfiguring the design process. It reviews the application of generative adversarial networks, diffusion models, and large language models in architectural planning, scheme layout, and structural optimization, and combines the practical cases of domestic architectural design enterprises from 2024 to 2026 to discover that the AI integration system based on BIM as the data hub is bridging the information gap between concept generation and construction drawing output. Generative AI enables space design to gradually break free from the limitations of standardized modular constraints, and user intentions can directly participate in the iterative generation of spatial schemes through natural language interaction. The promotion of technology also faces problems such as inconsistent data standards, insufficient algorithm interpretability, and a shortage of interdisciplinary talent. It can be promoted collaboratively from three aspects: institutional norms, technological research and development, and talent cultivation, to provide a reference for the intelligent transformation of the architectural design industry.