The API seems pretty simplistic. Does it choose the different supervised learning algorithm underneath without input from the user or does it just use one supervised learning algorithm? Their website does not give any algorithmic details.