Moving fluids at the micro- and nano-scales requires a different approach compared to the traditional methods based on pressure gradients. The increase of surface/volume ratio as the length of the systems decreases the efficiency of the latter. The use of thermodynamic forces such as electric fields, chemical potential and temperature gradients is crucial for effective transport when the local perturbation of the fluid at the interfaces becomes relevant. The specific case of chemical potential gradients in fast-moving components of a solution driving the movement of colloids, polymers and other moieties is known as diffusiophoresis. In this thesis, we study diffusiophoresis using a combination of theory and computer simulations. We use non-equilibrium thermodynamics, starting from the entropy production to build our theoretical framework aiming to discuss some subtleties present in previous works. As the first case of study, we perform non-equilibrium molecular dynamics to analyse the movement of a very large colloid using the Derjaguin-Anderson approximation, reducing the problem to a diffusio-osmotic flow. In the simulations, we drive the system out of equilibrium by applying microscopic representations of the chemical potential gradient that are compatible with periodic boundary conditions. We then report the first applications of such a method to a colloidal system and compare it with simulations where explicit concentration gradients are imposed. Our approach is more convenient as we decouple diffusiophoresis from strong advective effects. We find a non-monotonic relation between the diffusiophoretic mobility of the colloid and the strength of the interaction with the solution. Finally, we report a numerical study of polymer diffusiophoresis, finding that the existing theory for solid particles is not accurate for polymer coils. Moreover, we observe that the hydrodynamic flow through the polymer is less screened than for pressure-driven flows.