Abstract : An important class of visual analytics applications has to deal with dynamic data, which is continuously updated (e.g. by receiving new additions) while the analysis process is running. For instance, a visual analysis of Wikipedia articles, based on a live feed of Wikipedia data. The analysis has to gracefully adjust to the updates; stopping the process and triggering a total re-computation is not acceptable. We have designed ReaViz, a reactive workflow platform, conceived and deployed in close connection with a database of application and workflow-related data. ReaViz enables the declarative specification of data-driven workflows, enforcing a clean separation between the process data, the process structure, and the user interaction which may or may not involve visualization. A distinguishing feature of ReaViz is the possibility provided to the process designer to specify reactive behavior in the event of data changes to one or several of the data collections involved in a process. Thus, ReaViz aims at enabling the easy, elegant specification of visual analytics applications where data dynamics is an important aspect.