The client side of an Arbortext Publishing Engine publishing operation is handled by Java code called the Arbortext Publishing Engine Client Composer. The Client Composer collects all of the information that the Arbortext Publishing Engine Server Composer needs to perform some portion of a publishing operation, transmits the data to the server, awaits a response, and processes the data returned by the server. The server operations are explained in Arbortext Publishing Engine Server Composer.

The overall architecture of Arbortext Publishing Engine publishing—on both the client and the server—is designed to prevent content pipelines and content compilers from being aware of where they're running. Some content pipeline filters and some content compilers contain open source or commercial code that could not be modified to understand Arbortext Publishing Engine functionality.