The demands made on materials in contemporary design are increasingly stringent. Materials and design methods have evolved to meet them. However, it is proving increasingly difficult to extend current design methods, largely based on continuum modelling and empiricism, to cope with the larger number of variables that appear in many engineering applications. It is argued that atomistic modelling (which, by itself, seldom leads to engineering solutions) can give additional insight into the form of constitutive equations, the grouping of variables, and the magnitudes of material properties. Properly interpreted, this information can point the way to a `model-informed' empiricism that can help solve pressing engineering problems.