The challenge aim to solve the problem of predicting if a song will become a Billboard hit and build playlists around it using AI. The used dataset has almost 18 000 songs and was retrieved from Kaggle, which supports a variety of datasets publications formats. We tested seven different machine-learning algorithms to perform hit predictions, being Random Forest the most successful one, with 80% of average accuracy. As for the playlist building problem, we took advantages of K-Nearest Neighbours algorithm and the multi-criteria TOPSIS method to make selections.