Living systems are commonly engineered through an iterative design-build-test process. However, unlike in other engineering disciplines, the systems that synthetic biologists construct are able to evolve. On the one hand, this compromises their long-term stability. On the other, it opens up new avenues for the design of molecular tools or artificial cellular behavior through directed evolution. The fundamental advantage of directed evolution over rational bioengineering approaches is that the genotype-to-phenotype mapping can be navigated without being explicitly known. To improve the function of interest, it suffices to design a suitable selection pressure. Recent advances in in vivo selection mechanisms enabled continuous evolution of molecules. In these methods, the survival of the organism is tied to the performance of the protein of interest. Thus, all steps in directed evolution are performed by the host organism, increasing the throughput of the screens. Targets of previous continuous evolution campaigns have been single-conformation molecules. However, essential to biological computation and synthetic biology are proteins that can change activity in response to a signal from the environment. Directed evolution of such multi-state, dynamic targets is challenging, as it requires independent selection pressures that act on each of the protein's states and transitions between them. Hitherto, this has been achieved through frequent manual interventions. The main contribution of the thesis is the development of a continuous method for directed evolution of dynamical proteins and genetic networks. In the first project, the evolution of dynamic proteins is achieved by coupling them to a cell-cycle regulator that must oscillate for cellular self-replication. Since the cell-cycle gene is required, the system cannot be eliminated by mutations that deactivate it. Conversely, as the gene's expression must oscillate, it cannot be turned on constantly. Hence, using such a control, screens for both on and off states of the protein are performed once per cell cycle. Additionally, the selection pressure is encoded by light, which can be dynamically administered to cells over long periods. Light-directed evolution is applied to construct new chemogenetic and optogenetic regulators. We describe mutations in the El222/LIP transcription system that increase its sensitivity to light, decrease its leakiness or change its spectral responsiveness. For the PhyB/Pif3 optogenetic switch, we uncovered a mutation that renders the exogenous supplementation of the chromophore unnecessary. Finally, we evolve variants of the Tet-On system that are more sensitive to doxycycline. The second project focused on the application of light-directed evolution to study the rules that govern the timing of cellular replication. Despite the diversity in microbial replication times, how different cell cycle timings have evolved is not understood. By forcing budding yeast to divide faster at the expense of other traits for hundreds of generations, we sped up the cell cycle by around 10%. The identified phenotypes will provide insight into the benefits of slower cellular replication. Lastly, I describe several contributions that focused on developing tools for accurate and generalizable segmentation of microscopies, a tool for automated design of sequences for genome editing, and a benchmark of systems for transcriptional control in budding yeast.