LLM-Powered Micro Well-being Habit Interface in Personal Task Management (PTM): Promoting Positive Self-Reflection

Abstract

This study investigates the integration of Large Language Model(LLM)-powered wellbeing micro break habit recommendations into Personal Task Management(PTM) systems to promote positive self-reflection among young adults. Despite the widespread adoption of PTM applications, users experience a frequently intention-behavior gap, failing to translate their goals into actionable habits. This research examines micro break habits that provide specific, short-duration actions (within 5 minutes) with behavioral cues. To address this challenge, we developed and evaluated a system that leverages LLM technology to recommend personalized micro break habits and facilitate automated self-reflection. Through a 28-day user study with six long-term PTM users, we identified key behavioral characteristics including dynamic priority shifts, contextual relationships between tasks, and the correlation between mid-priority behaviors and user satisfaction. Based on these insights, we developed "Tiny Well-flect," a web application featuring routine-based well-being habit recommendations, AI-generated reflective journaling, and progress visualization. The system achieved a System Usability Scale(SUS) score of 71.8 with 206 participants, and 82.5% responded positively to the automated journaling feature. This research demonstrates how LLM technology can bridge the intention-behavior gap in PTM systems by delivering personalized micro break recommendations that provide contextually relevant wellbeing activities and generating reflective journals that reinforce well-being lifestyles.