We report on a type-theoretic method for functional synthesis of processes from repositories of components. Our method relies on the existing framework for composition synthesis based on combinatory logic, (CL)S. Simple types for BPMN 2.0 components and a taxonomy of domain specific concepts are used to assign types to BPMN 2.0 fragments and functional fragment constructors. Both serve as input for the automatic creation of meaningful processes. Staging synthesis into two levels provides a separation of concerns between the easy task of extracting fragments from existing processes and the more sophisticated task of deducing functional fragment transformations. We study the applicability of the described approach by synthesizing control processes for LEGO® Mindstorms® NXT robots deployed on the Activiti platform. We evaluate experimental results analyzing synthesized processes regarding correctness, variability and the time consumed for their creation by the (CL)S framework. Additionally, the steps necessary to target a different application domain are described.