concept: an infinite sound library for people who make music
everything is generated on the fly with LLMs using implicit prompting from navigation. ai music tools should be more like this — augmenting workflows, not replacing artists
note: this is just a clunky proof of concept with no audio