Evolutionary scientific problems usually have to be addressed indirectly, by comparing observations in extant species and inferring what happened millions of years ago. While this approach has taught us a lot and provided many hypotheses, it remains difficult to test these ideas. By applying phylogenetic analysis it is possible to statistically estimate what proteins looked like millions of years ago. The best estimate can then be studied by resurrecting the protein through gene synthesis and functionally studied. We thus far applied this strategy to three problems in evolutionary biology: the origin of flowers, the impact of whole genome duplication and the origin of auxin and jasmonate signalling. For the origin of flowering plants, we provide evidence that current floral quartets specifying male organ identity, which consist of four types of subunits, evolved from ancestral complexes of two types of subunits through gene duplication and integration of SEPALLATA proteins just before the origin of flowering plants. For the impact of ancient whole genome duplication, we found that such an event can strongly rewire protein interaction networks which allows for the origin of new functions. For the origin of auxin and jasmonate receptor specificity, our results suggest that the ancestral receptor was not functional yet and evolved independently into an auxin and jasmonate receptor. The auxin receptor appears to have evolved more rapidly and was already functional at the origin of land plants. Together, these examples illustrate the power of this approach to make hypotheses in evolutionary biology testable.