Conversation in the cloud is a live multimedia composition for two human musicians and one AI musician. The conversation between three musicians starts at the intersection of reality and virtuality via music. Then, musicians will drive their limits through improvisation founded on each other’s music during the live performance. Finally, the two worlds slowly merge as the conversation deepens. The AI musician is a comprehensive system that involves multiple machine learning techniques to enhance its machine musicianship, such as deep neural networks and human body pose estimation. Thus, the combination of live multimedia and performances from three musicians unveils a multidimensional music conversation.