This study introduces the development of a platform for process-based assessment in music performance and its application in schools. The platform was designed to support four performance activities that are explicitly presented in the Korean national curriculum and are most widely practiced in schools: Western vocal music, Korean traditional vocal music, recorder, and danso (a Korean traditional wind instrument). The project sought to provide a genuine case of process-based assessment in practical music classrooms following curriculum standards (Ministry of Education, 2015; MOE, 2022).Information about the basic features needed was obtained through surveys and interviews with secondary music teachers. They agreed that process-based assessment is valuable but said it is hard to manage in large classes where it’s not easy to give individual and continuous feedback. They pointed out several needs: automated feedback on pitch and rhythm, a place where students’ practice data can be stored and reviewed, and functions that allow qualitative teacher feedback to be integrated with quantitative results (Kang, Shin, & Jung, 2025). Based on these insights, the platform was developed to review students’ performances in real time, with instant feedback on pitch and rhythm correctness in addition to AI-generated qualitative feedback on their work. In addition, teachers can later listen back to the recordings and provide feedback, enabling a balanced approach that combines automated responses with teacher-led assessment.The platform was piloted in school contexts to examine its educational applicability. Students responded positively to the immediate feedback, which made them practice more actively. Teachers said the stored data helped them follow student growth without losing track. However, real classrooms turned out to be complicated. Recognition accuracy varied depending on internet connection and computer conditions. For instruments such as recorder and danso, recordings made through microphones were often unclear, picking up background noise or even the sounds of nearby players.The findings show that using AI in music assessment has clear benefits but also some real limits in practice. The platform supported students’ learning and teachers’ assessment practices, but further technical refinement and broader classroom testing are still required to ensure stability and flexibility in diverse educational contexts.