Solvents are of great importance in many technological applications, but are difficult to study using standard, off-the-shelf ab initio electronic structure methods. This is because a single configuration of molecular positions in the solvent (a "snapshot" of the fluid) is not necessarily representative of the thermodynamic average. To obtain any thermodynamic averages (e.g. free energies), the phase space of the solvent must be sampled, typically using molecular dynamics. This greatly increases the computational cost involved in studying solvated systems. Joint density-functional theory has made its mark by being a computationally efficient yet rigorous theory by which to study solvation. It replaces the need for thermodynamic sampling with an effective continuum description of the solvent environment that is in-principle exact, computationally efficient and intuitive (easier to interpret). It has been very successful in aqueous systems, with potential applications in (among others) energy materials discovery, catalysis and surface science. In this dissertation, we develop accurate and fast joint density functional theories for complex, non-aqueous solvent enviroments, including organic solvents and room temperature ionic liquids, as well as new methods for calculating electron excitation spectra in such systems. These theories are then applied to a range of physical problems, from dendrite formation in lithium-metal batteries to the optical spectra of solvated ions.