When metal nanoparticles are deposited on a reducible oxide support, an unique synergy appears at the boundaries. Downscaling the size to that of a single atom has become a hot topic of late since the chemistry alters substantially. The first half of this thesis focuses on platinum atop ceria, which is a key component in three-way car-exhaust catalysts. As the US Department of Energy is looking for greener, more fuel-efficient combustion engines, exhaust temperatures have to decrease significantly as well. Experimental findings suggest that single-atom platinum can achieve these lofty ambitions. I will present the first-ever DFT verification of these observations along with the mechanism of dynamic oxidation states behind it. Its scope appears to be broader than just platinum and the discussion readily extends to the whole of group 10. Nevertheless, even state-of-the-art computational methods have their restrictions. In the second half I will be covering two proofs of concepts for innovation. On the one hand we are collaborating with VASP developers in the benchmarking of a yet unreleased, low-scaling MP2 implementation that should improve accuracy. On the other, we are breaking the scale limitations inherent to ab initio using Artificial Intelligence to tailor a forcefield specifically to ceria.