CC BY 4.0Hsu, Shu ChienLee, Hui-Chu2026-08-142026-08-142026-08-14https://2026.ifla.orghttps://repository.ifla.org/handle/20.500.14598/7238To address rapid advances in AI, Kaohsiung Medical University Library launched an AI-powered Virtual Counter Service in June 2023, integrating ChatGPT to provide both text and voice consultations. This study examines user acceptance and experience to inform AI library development. A mixed-methods approach was employed. Surveys conducted during Library Week in 2023 and 2024 yielded 452 and 473 valid responses, respectively, while in-depth interviews with five frequent users were conducted in 2025. Results indicate that satisfaction with convenience, efficiency, and response accuracy exceeded 94%. The system evolved from a text-based interface to an interactive model with voice recognition. Ongoing enhancements include expanding the Knowledge Base, improving response accuracy, optimizing the interface, implementing feedback mechanisms, and strengthening book search functions. Agent AI will be further integrated to enhance proactive and context-aware services. Findings reveal strong user acceptance and high intention to use. Integrating survey and interview results not only enhances service awareness but also strengthens the library’s innovative and intelligent image. As AI Agents technologies continue to advance, user-centered design will further improve experiences and drive digital transformation in libraries.enghttps://creativecommons.org/licenses/by/4.0/Academic and research librariesArtificial intelligenceReference servicesUser experiencePostersHsu, Shu ChienLee, Hui-Chu