A network system for enabling voice interaction between communications-center applications and human agents remote from the center has a primary server connected to the network the server controlling at least one routing point used by the center, a secondary server connected to the network the secondary server for generating and serving voice extensible markup language, a voice gateway associated with the secondary server, the gateway for executing voice extensible markup language and recognizing speech input, and a software platform based in the primary server and distributed in part as a server application to the secondary server, the software suite functioning as a data transformation interface between the center applications and the gateway. In a preferred use agents and applications communicate bi-directionally using VXML.