Proactive Conversational AI for Enhanced Mobile Automation Dexterity

This paper introduces a novel paradigm in mobile interface automation, emphasizing autonomous agent-initiated dialogue. We address the critical challenge of seamlessly integrating user clarification within automated task execution on smartphone UIs. By enabling intelligent agents to proactively identify and resolve ambiguities through natural language interaction, our approach significantly improves task completion rates and user trust. This work lays the groundwork for more adaptable and user-aware automation systems, moving beyond static action planning to dynamic, human-like collaborative mobile task fulfillment.