Once lost in the wonderland of machine learning, our mind sometimes begins dreaming of self-determining applications capable of reading lips, scan bodies and, especially, control humans. The reality is that machine learning is only a bunch of smart mathematics and all it can do is getting numbers and returning numbers. Plain software applications are then responsible for using and consuming those numbers in possibly scary ways. In this session, we’ll demystify the alleged magic behind machine learning and, using linear regression as our driving vector, we’ll show what happens in the training phase, what happens in production and what’s ultimately stored in the trained model.