This paper proposes a new method for programming and controlling distributed tasks. Applications declare behavior patterns that are used to automatically predict and reserve resources needed by applications in a heterogeneous distributed environment. The paper introduces a stochastic quality sampling driven scheduler and a rendezvous mechanism for matching pre-planned activities with actual payload data. The system is built around the first real-time SOAP implementation, also presented in this paper. It extends the XML Web Services interoperability benefits that have proven themselves in e-business into two new areas: embedded and real-time. The paper presents an efficient implementation that runs on common microcontrollers and other computers.