The ability to manipulate microscale objects with precision to form complex structures is central to the field of micro-robotics and to the realization of reconfigurable systems. Understanding and exploiting the forces that dominate at the microscale in complex environments pose major challenges and open untapped opportunities. This is particularly the case for micro-particles in soft milieu like fluid interfaces or nematic liquid crystalline fluids, which deform or reorganize around dispersed colloids or near bounding surfaces. These energetically costly deformations can be designed as embedded energy landscapes, a form of physical intelligence, to dictate emergent colloidal interactions. The fluid nature of these soft milieu allows colloids to move to minimize the free energy and externally forced robotic structures to re-write the embedded energy landscapes in the domain. Such physically intelligent systems are of great interest at the intersection of materials science and micro-robotics. Micro-particles on fluid interfaces deform the interface shape, migrate, and assemble to minimize the capillary energy. In the first part of my thesis, I design and fabricate a magnetic micro-robot as a mobile curvature source to interact with passive colloids on the water/oil interface. An analytical expression that includes both capillary and hydrodynamic interactions is derived and captures the main feature of experimental observations. I further demonstrate multiple micro-robotic tasks including directed assembly, cargo carrying, desired release and cargo delivery on the interface. Micro-particles in confined nematic liquid crystals (NLCs) distort the nematic director field, generating interactions. These interactions depend strongly on the colloids shape and surface chemistry, geometric frustration of director field and behavior of dynamic topological defects. To probe far-from-equilibrium dynamics, I fabricate a magnetic disk with hybrid anchoring. Upon controlled rotation, the disk’s companion defect undergoes periodic rearrangement, executing a complex swim stroke that propels disk translation. I study this new swimming modality in both high and low Ericksen number regimes. At high rotation rates, the defect elongates significantly adjacent to the disk, generating broken symmetries that allow steering of the disk. This ability is exploited in path planning. Thereafter, I design a four-armed micro-robot as a mobile distortion source to promote passive colloids assembly at particular sites via emergent interactions in NLCs whose strengths are characterized and found to be several orders of magnitude larger than thermal energies. While the strength of theses interactions allows colloidal cargo to be carried with the micro-robot during translation, it poses challenges for cargo release. We find that rotation of this micro-robot generates a complex dynamic defect-sharing event with colloidal cargo that spurs cargo release. Thereafter, I demonstrate the ability to exploit NLC elastodynamics to construct reconfigurable colloidal structures in a micro-robotics platform. At the colloidal scale, rotation dynamics are easier to generate, and this motivated me to exploit the topological swimming modality of the micro-robot. Using programmable rotating fields to direct the micro-robot’s motion, I achieve fully autonomous cargo manipulations including approach, assembly, transport and release. The ability to dynamically manipulate micro-particles and their structures in soft matter systems with embedded energy landscapes, as demonstrated in this thesis, creates new possibilities for micro-robotics and reconfigurable systems.