Using cognitive modeling to predict user behavior in the human-computer interaction domain is a promising field, but often hindered by practical problems. Especially the creation of a mock-up of the technical system under evaluation is often a tedious and time-consuming task. For the growing number of HTML-based applications, the modeling toolbox ACT-CV (Halbrügge, 2013) provides direct access to the user interface, removing the necessity to create a mock-up before the actual modeling can start. A down-side of this approach is that unaltered applications often cannot be used for fast-time simulation. This paper presents a new tool that solves this problem by capturing UI states and also the control flow of HTML applications and by transforming both into finite automata that can be used for fast-time simulation.