This study focuses on the development of JAAM Music, a web-based intelligent learning system designed to help individuals learn string instruments such as the bandurria, acoustic guitar, electric guitar, ukulele, and bass. The goal is to offer a more accessible and engaging music learning experience by combining interactive tutorials, gamified activities, real-time AI feedback, and progress tracking. The system is being developed using the Agile methodology to support continuous improvement throughout the process. At its core, JAAM Music uses a hybrid Convolutional Neural Network (CNN) and Transformer model to recognize sounds and provide real-time feedback based on the learner’s audio input. Video lessons and gamified activities are also included to keep users motivated. We also worked closely with fellow musicians and music educators to design the system in a way that addresses common challenges faced by learners. Preliminary findings based on early evaluations from musicians and professionals suggest that the system has strong potential to improve motivation, skill acquisition, and overall learning outcomes. The AI-driven feedback aims to simulate instructor guidance, helping users improve as they practice. Future work will involve full user testing, system evaluation, and further enhancements to the AI model to better support learning goals.