Objective: Non-pharmacological interventions are considered the first-line treatment for behavioral and psychological symptoms of dementia (BPSD); however, traditional approaches have shown only small effect sizes. Mobile technology offers an opportunity to improve BPSD assessment and management in people living with dementia (PLWD). We aimed (1) to develop a mobile application (app) featuring a real-time BPSD diary, machine-learning-based BPSD prediction, and individualized non-pharmacological care programs, including therapeutic use of music and reminiscent content, and (2) to test its usability, acceptability, and preliminary efficacy among PLWD and caregivers. Methods: An Android-based app was developed through the following three phases: (1) needs assessment, (2) software development and initial testing with experts, and (3) beta-testing with end users who were dyads of PLWD and caregivers. The preliminary efficacy, usability, and acceptability of the app were assessed using validated BPSD questionnaires and face-to-face interviews with the dyads. Logs of the dyads’ program participation (i.e., types, time, and duration), BPSD diaries, and engagement levels of PLWD were also collected through the app. Results: Five dyads created BPSD diaries (range: 22–48) over 3 weeks. Overall, the BPSD symptoms decreased after the beta-testing period. Each dyad participated in the care programs for 106–204 min, during which music alone was most frequently used. Engagement levels ranged from 3.38 to 4.94 (out of 5). Conclusions: The app was deemed usable, acceptable, and feasible for PLWD and caregivers. The upgraded app will be further tested and can be easily implemented at home or in the community.