String theory phenomenology generically suffers from either too much flexibility (and lack of predictability) or from the a high specialization to case by case studies. I will discuss how F-theory GUT model building manages to get around these pitfall, in particular, I will explain, how to systematically include global string consistency conditions, which are independent of the specific compactification, and which come with the benefit of highly constraining the class of GUT models that can arise from F-theory.