This research is motivated by two significant challenges facing the planet: reducing the emission of CO2 to the atmosphere and production of sustainable fuels by harnessing solar energy. The main objective of this work is the study of promising photocatalysts for CO2 reduction. DFT modeling of CO2, subnanometer Ag&Pt clusters, and anatase TiO2 (101) surface is employed to gain fundamental understanding of the catalytic process, followed by validation using a guided experimental endeavor. The binding mechanism of CO2 on the surface is investigated in detail to gain insights into the catalytic activity and to assist with characterizing the photocatalyst. For CO2 photoreduction, the cluster induced sub-bandgap and the preferred adsorbate in the first and key step of the CO2 photoreduction are explored. It is found that TiO2-supported Pt octamers offer key advantages for CO2 photoreduction: 1. by providing additional stable adsorption sites for favored CO2 species in the first step, and 2. by aiding in CO2- anion formation. Electronic structure analysis suggests these factors arise primarily from the hybridization of the bonding molecular orbitals of CO2 with d orbitals of the Pt atoms. Also, structural fluxionality is quantified to investigate geometry dependent (3D-2D) CO2 adsorption. Geometric information, electronic information, and C-O bond breaking tendency of adsorbed CO2 species are proposed to connect to experimental observables (IR frequency). The CO2 adsorption sites on supported Pt clusters are also identified using IR as the indicator. A cluster-induced CO2 dissociation to CO pathway is also discovered. Finally, experimental work including dendrimer-encapsulated technique, TPD, and UV-Vis is performed to validate the computational results, the availability of adsorption sites and CO2 binding strength on supported Pt clusters.