In the Define phase, our primary goal was to clearly identify the problem, our target user, and establish a focused direction for our solution. After analyzing our interviews, we synthesized our findings into a clear problem statement: Help college students and young adults manage their busy or relaxed schedules in a way that caters to their individual style, while considering self-care and socialization. Rather than simply building another calendar app, we defined our opportunity as creating a system that reduces burnout through personalized schedule optimization.

To better understand our target audience, we developed detailed persona maps. Our primary persona was a college student balancing academics, social commitments, part-time job, and personal responsibilities. In building the persona, we included demographic context, goals, needs , and challenges and frustration. We grounded the persona in patterns identified through the data we gathered in our interviews. This helped us access the problem and ensure that design decisions remained user-centered.

Next, we created user journey maps to visualize the students’ experience across time. We mapped the stages from receiving assignments, planning tasks, managing weekly schedules, experiencing deadline pressure, and reflecting afterward. For each stage, we identified user actions, thoughts, emotions, and friction points. For example, during the planning stage, students often underestimate task duration. During the execution stage, tasks are frequently rescheduled multiple times. Near deadlines, stress spikes and sleep is compromised. Mapping this journey allowed us to pinpoint intervention opportunities. The journey map made it clear that our solution needed to be proactive, not reactive, and capable of learning from repeated behaviors.

As we refined the solution concept, we identified where AI would provide meaningful value. We determined that AI was most appropriate for analyzing schedule patterns, detecting avoidance signals, evaluating cognitive load across time blocks, and generating adaptive recommendations for studying, sleep, eating, and rest.

Overall, we defined our AI data needs and mapped each required data element directly to a user need. To generate realistic schedule recommendations, we identified the need for event type, start and end times, non-adjustable responsibilities, and adjustable tasks. To prevent burnout, we included cognitive load ratings and rest goals. To preserve social balance, we identified preferred social windows and optional friends’ schedule data. To enable system learning, we defined key behavioral labels such as accept/rejection of recommendations, user feedback on rejected suggestions, tasks rescheduled more than twice, and tasks completed close to deadlines. By clearly connecting each data requirement to a specific user outcome, we ensured our data remained purposeful and aligned with reducing burnout.