video1.0<iframe src="https://www.loom.com/embed/867a1cea0d5647d38bb3d1bcf8830fd9" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/sessions/thumbnails/867a1cea0d5647d38bb3d1bcf8830fd9-542f3b10b493df7a.gif300.826I built a mood based music recommendation app that calls the Cloud API and uses RAG, then takes a user prompt like something chill or something mellow. The system retrieves songs and runs confidence scoring using energy, valence, acousticness, and BPM, reporting metrics like tempo 152 at 0.93 and 132 in another run, with confidence scores such as 0.74 pass and 0.68 pass. I added guardrails, so if it does not pass, it gives different results. I also ran a reliability check with six Claude queries, saving results to JSON, with six passed and average confidence 0.73. I did not request any viewer action, but I noted a future improvement to add the Spotify API.