Music recommendation is a solved problem, if your goal is keeping someone on the app. Spotify knows exactly what to play to make sure you never leave.

GrooveFind is built around a different question: what would you find if the algorithm wasn't trying to keep you?

Type anything. The color blue at night, the feeling of a long drive through the desert, a supporting character from a book you read in high school. What does that idea actually sound like? GrooveFind exists to answer that question.

Under the hood, the input goes through an LLM call that extracts musical DNA (texture, mood, tempo, energy, structural feel). That gets mapped to SoundCloud's catalog through a series of targeted searches, then filtered through an algorithm I developed to model how people actually make judgments about musical fit.

Forget time-on-app, skip rate, or re-engagement. The only question I care about is did you like your playlist?