Background <p>While physical activity (PA) has protective effects in mitigating dementia progression among individuals with mild cognitive impairment (MCI), suboptimal exercise adherence remains a critical barrier. Mobile health technology, when integrated with preference-based personalization strategies, may enhance adherence through tailored interventions. However, empirical evidence validating the feasibility of exercise preference-driven mobile solutions for populations with MCI remains limited.</p> Objective <p>To develop and evaluate a mobile exercise management application specifically designed for community-dwelling older adults with MCI that incorporates personalized preference-driven features.</p> Methods <p>A three-phase mixed-methods study was conducted: (1) Formative research: key design elements of the exercise application were identified based on evidence from existing interventions and users’ exercise preferences (self-selected by participants). (2) Prototype design and development: an iterative methodology was employed to design, develop, and test the application’s functionality. (3) Usability evaluation: A 4-week usability evaluation was conducted with community-dwelling older adults with MCI using the application.</p> Results <p>Twelve participants (age = 70.60 ± 3.21; MoCA = 21.67 ± 2.23) completed the 4-week usability evaluation. The application demonstrated good usability (SUS = 77.73 ± 6.73), with optimization targets identified in functional integration (item 5 converted score = 2.67 ± 0.49) and initial technical support needs (item 7 converted score = 2.33 ± 0.49). The median exercise adherence (exercise goal completion rate) among participants was 100.00% (87.48, 100.00), with 75% (9/12) of participants achieving or exceeding their weekly goals. The exercise preference agreement rates exhibited considerable variability: environment—100.00% (93.55, 100.00) and modality—97.83% (63.40, 100.00). While there was a strong positive correlation between the environment and modality preference agreement rates (Spearman rho = 0.817 [<i>p</i> = 0.001]), there was no association between preference agreement rates and adherence (<i>p</i> &gt; 0.05). Users highly valued the application’s real-time monitoring, motivational features, and intuitive interface, with most reporting no significant usability issues.</p> Conclusions <p>This 4-week, single arm usability study demonstrates that preference-driven mobile exercise management is not only feasible but also holds promise for promoting exercise engagement in community-dwelling individuals with MCI. User feedback identified critical optimization needs in exercise type diversity and initial technical support. While synergistic relationships between exercise environment and modality preference agreement rates were evident, the absence of a correlation between preference agreement rates and exercise adherence underscores the need for longitudinal studies to explore dynamic preference-behavior interactions. Technology adoption barriers (e.g., self-efficacy, digital literacy) and memory-associated challenges necessitate multimodal objective monitoring. These findings highlight two implementation priorities: prioritizing accessibility in exercise environments and incorporating personalized adaptation strategies to address exercise pattern variability.</p>