Antiretroviral therapy (ART) is directly responsible for reducing the death rate from HIV/AIDS. AIDS-related deaths declined by 38% from 2000 to 2017—saving more than 11 million lives worldwide. Mediation adherence is critically important to this statistic and to clinical trials. Investigators insist on (or assume) adherence rates of ≥95%. However, when bodily fluid measurements are used to assess adherence, the values are substantially lower: 54% to 68%. Viral suppression generally requires most ART be taken with at least 95% adherence. Poor adherence to ART during a clinical trial can underestimate efficacy, increase subject discontinuation rates, and extend recruitment periods and the total duration of clinical trials. This is expensive for clinical trial sponsors and slows or halts the development of antiretrovirals. Direct observation is the gold standard for medication adherence but is prohibitively expensive and impractical. HiDO is an automated AI-driven direct observation medication adherence platform. The platform is a 510K-exempt, Class I medical device with a provisional patent that integrates medication dispensing, pill count and a front-facing video cameras to confirm the right medications are given at right time to the right patient. Investigators have access to video observation logs, patient dose time, adherence trends, and study-level adherence through the platform’s dashboard. Data is stored securely in the cloud and accessible real-time. The device dispenses up to 7 different types of medications simultaneously, 40 doses each. During the 90-day study, participants will receive smartphone reminders; pills will be dispensed through the unit and adherence monitored and verified through video observation and facial recognition. Our benchmark for success is that all participants who complete the study will achieve ≥95% adherence to ART averaged across 90 days (Milestone 1). Adherence will be quantified as the (no. of doses provided — no. of doses taken) / the no. of doses provided X 100. We will further confirm adherence by reviewing patient charts for viral load and CD4 T-cell count. In the same study population described above, we will conduct full usability testing at 2 weeks of use. We will measure Time on Task for initial registration, “first click” testing, facial recognition setup, and medication administration, and assess the number and type of critical and non-critical errors and error-free rate. Results of usability testing will be equal to or better than published benchmark rates for similar platforms (Milestone 2). Subjects will complete the 10-item System Usability Scale (SUS) and achieve an average SUS score >68 (Milestone 3). More than 80% of subjects will have Likely or Strongly Likely Net Promoter Score (Milestone 4). If successful, we will have demonstrated that our automated medication adherence platform is highly usable and user-friendly, enables strict adherence to ART trials, and is ready for testing in Phase ...