Macroscopic robotics has become an integral part of our daily lives, inspiring researchers to develop smart nanoparticles to initiate a similar revolution on the microscale. In this study, we employ molecular dynamics simulations to explore the robotic capabilities of "flexicles"—a computational model of a three-dimensional, deformable, and hierarchical object whose motion is dictated by self-propelled colloids confined within a flexible vesicle. By placing flexicles in various environments and scenarios, we demonstrate their ability to autonomously perform complex tasks such as curvature-dependent geometrical trapping, navigating topographical obstacles, and transporting passive cargo particles. We identify a physical mechanism underpinning these behaviors, linking the interactions between flexicles and curved surfaces to deformations that influence the arrangement and ordering of internal active colloids. Our results establish flexicles as promising model candidates for dynamic, active, cell-like, liquid matter microrobots and provide a foundation for developing physics-based strategies for other autonomous microrobotic behaviors in the future.