Recent advances in artificial intelligence (AI) are catalysing new paradigms in music education. This paper presents the design and implementation of Darcy, an AI tutor built on large language models (LLMs), within a distance-learning music technology course at the University of Brasília. Adopting an applied research approach, the study combined system development with qualitative evaluation through semi-structured interviews. Darcy was developed as a 24/7 support tool to assist students with both technical issues and music-related content, drawing on course materials and frequently asked questions.Three experienced tutors interacted with the prototype and provided evaluative feedback. Results indicate that Darcy effectively responded to common queries, delivered personalized guidance, and supported learner autonomy through immediate feedback. Tutors acknowledged the AI's potential to enhance educational experiences—clarifying concepts, recommending exercises, and adapting responses to individual needs—while emphasizing that such tools should complement, rather than replace, human instruction.Challenges identified include ensuring content accuracy, fostering student engagement, improving mobile accessibility, and addressing ethical concerns such as data privacy and algorithmic bias. The findings suggest that AI tutors can function as collaborative agents in music education, fostering interactive, personalized, and adaptive learning environments. The study underscores the importance of multidisciplinary collaboration, clearly defined usage boundaries, teacher professional development, and ongoing refinement to align AI applications with pedagogical goals and ethical standards.