BPEL can automate orchestrations for cross-organizational Web services; however, it meets a serious challenge from modeling human-intensive business activities. This paper distinguishes human-driven workflow from automated process, and uses state machine to model human-driven workflow, then transforms state machine-style workflow to BPEL process, finally integrates BPEL-style human-driven process with automated process. The goal of our work is to utilize the flexibility and efficiency of state machine to model complex and dynamic business process, and provides a novel procedure to model human-intensive business activities to complement the limitation of BPEL. This procedure has been implemented in DartFlow - a service composition platform for the sharing of the knowledge and services.