Data science projects often have to overcome two opposite, yet equally damaging, preconceptions. On one hand, the hype around AI/ML leads some researchers to think that it can magically solve any question, regardless of the problem and availability of data. On the other hand, skeptical users perceive AI/ML models as black boxes, and consider their predictions not reliable. We aim to find the middle ground, increasing the adoption of AI/ML thanks to use-case examples, strict open science guidelines and full documentation of each model, including training set, possible biases, application domain etc.