PersonalAlign: Hierarchical Implicit Intent Alignment for Personalized GUI Agent with Long-Term User-Centric Records

Real-world deployment of GUI agents requires aligning with users' complex implicit intents, beyond explicit instructions. This paper introduces "PersonalAlign," a new agent task where agents utilize long-term user records as persistent context to resolve omitted preferences in vague instructions and anticipate latent routines for proactive assistance. The AndroidIntent benchmark is proposed for this task, and the Hierarchical Intent Memory Agent (HIM-Agent) is introduced. HIM-Agent maintains a continuously updating personal memory, hierarchically organizes user preferences and routines, and significantly improves execution and proactive performance.