"Having Lunch Now": Understanding How Users Engage with a Proactive Agent for Daily Planning and Self-Reflection

TL;DR

This paper investigates how users interact with a proactive coaching-style agent, PITCH, in daily planning and evening self-reflection through a 14-day field study with 12 graduate students. Using thematic and dialogue-act analyses of 336 conversations (3,181 turns), it reveals that users frequently externalize plans, report progress, and negotiate or resist agent suggestions, with notable differences between morning planning and evening reflection. The study identifies design implications for proactive coaches, including leveraging explicit agendas, memory to support external logging, and flexible pacing to avoid rigidity and over-prompting, while also noting breakdown patterns such as misaligned goals, excessive responsiveness, and meta-awareness gaps. A released anonymized dataset and actionable guidance contribute to the development of more effective, behavior-change-oriented conversational agents. The findings advance understanding of how AI-driven coaches can support personal productivity and well-being, emphasizing social dynamics, adaptability, and transparent capability communication as key design axes.

Abstract

Conversational agents have been studied as tools to scaffold planning and self-reflection for productivity and well-being. While prior work has demonstrated positive outcomes, we still lack a clear understanding of what drives these results and how users behave and communicate with agents that act as coaches rather than assistants. Such understanding is critical for designing interactions in which agents foster meaningful behavioral change. We conducted a 14-day longitudinal study with 12 participants using a proactive agent that initiated regular check-ins to support daily planning and reflection. Our findings reveal diverse interaction patterns: participants accepted or negotiated suggestions, developed shared mental models, reported progress, and at times resisted or disengaged. We also identified problematic aspects of the agent's behavior, including rigidity, premature turn-taking, and overpromising. Our work contributes to understanding how people interact with a proactive, coach-like agent and offers design considerations for facilitating effective behavioral change.

Figures (1)

• Figure 1: An example of the morning and evening conversations from the field study. PITCH checks in with the user in the morning to externalize their plan as shown in (a) and follows up with them in the evening given the morning conversation context as shown in (b).