FRAC0105 | User Acceptance Testing of Self-Cav: An AI-driven Digital Conversational Agent for Young South Africans

Paul Potsane, Ntombifikile Mtshali, Gertrude Mbhalati, Cleopatra Sokhela | Shout-it-now, Centurion, South Africa

This study aimed to assess the user acceptance of Self-Cav, a digital self-care conversational agent. The name “Self-Cav,” a local South African term for self-awareness, was intentionally chosen to resonate with young people, empowering them to understand their HIV and overall well-being needs. Researchers employed a mixed-methods approach. For feasibility, a Self-Cav prototype was developed, integrating evidence-based behaviour change techniques and content tailored for South Africa. A pilot conducted between May 29 and June 12, 2025, monitored technical functionality, user engagement, and system stability. For acceptability, a pilot evaluation involved 123 young people (18-35 years old) from diverse backgrounds, recruited via online and in-person QR codes and social media. Acceptability was measured quantitatively for perceived usefulness and ease of use, and qualitatively through group discussions. The Self-Cav agent demonstrated high technical functionality and stability, with 100% uptime and no reported bugs during the pilot. Of the 123 participants, 96 (78.0%) accepted the terms, and 53 (43.1%) actively engaged with the chatbot. Acceptability findings were overwhelmingly positive: 85% found SelfCav easy to use, and 78% perceived it as useful for supporting HIV-related behaviour change and well-being. Qualitative data from 60 participants highlighted the agent’s accessibility, non-judgmental nature, and discrete information delivery as key appealing factors, especially its connection to “self-awareness”. Self-Cav is feasible and highly acceptable to young South Africans for supporting HIV behaviour change and overall well-being. The positive reception of its culturally relevant name suggests enhanced relevance and uptake. These results underscore the significant potential of culturally resonant, AI-driven digital health interventions to effectively complement existing HIV response efforts, particularly in resource-constrained settings. The next phase will involve a larger sample size to further validate these findings.