Design Thinking–Driven Multi-Agent Method for Persona Generation
Qing Ao, Suihuai Yu, Jianjie Chu, Chaojie FangPersonas are widely used in user-centered design, yet conventional persona construction remains labor-intensive, difficult to update, and weak in evidence traceability. This study proposes a Design Thinking–driven multi-agent method for persona generation, organizing the Empathize, Define, Ideate, Prototype, and Test stages into a closed-loop workflow. Within this workflow, raw research materials are progressively transformed into structured evidence, requirement hypotheses, persona drafts, visual concept prompts, and validation feedback, enabling iterative refinement across stages. A prototype system was developed to implement the method and preserve stage-wise artifacts, improving traceability, interpretability, and versioned iteration throughout the persona development process. Initial application and evaluation results suggest that the proposed method can support scenario-grounded persona generation and early concept framing for human-centered design practice across collaborative design settings, while future work should further examine broader design contexts and stronger downstream specification support.