This thesis focuses on the development and study of artificial microscopic and nanoscopic active matter systems. In particular, it demonstrates that passive building blocks such as colloids can self-assemble into active molecules, engines and active droplets that display a rich set of active motions. This is achieved by combining optical manipulation with a phase-separating environment consisting of a critical binary mixture. The first part of this thesis shows how simple absorbing particles are transformed into fast rotating microengines and nanoengines using optical tweezers. The second part of this thesis demonstrates the self-assembly of colloidal molecules, which exhibit diverse behaviour such as propulsion, rotation and spinning, and whose dynamics are characterised by a novel machine-learning algorithm. Then, the interaction of colloidal molecules with their phase-separating environment is investigated and a two-way coupling between the induced liquid droplets and their immersed colloids is observed. At last, an alternative mechanism for the prevention of stiction in MEMS is demonstrated using fluctuation-induced forces. The insights gained from this research mark the path towards a new generation of design principles, e.g., for the construction of flexible micromotors, tunable micromembranes and drug delivery in health care applications.