Collaborative technologies can help improve work conditions for operators with any profile, in demanding production tasks. This paper presents an intervention in an intermediate stage of a Metal Injection Molding (MIM) manufacturing process with poor ergonomics, both physical and stress-related, due to sustained visual attention demands. A collaborative workstation was designed and implemented, integrating a collaborative robot and a machine-vision system for bin-picking. The strategy of the intervention was to distribute tasks between worker and robot, where the robot undertook the non-ergonomic task of arranging small pieces close together on a plate, forming a pre-established pattern. Tests with a first implementation suggest that the task-sharing intervention for the manipulation of small MIM-produced references is viable. Careful design of robot fingers is key for the manipulation of such parts, and ambiguous piece configurations require fine tuning of visual piece identification systems, for error-free execution.