Building and using agent-based models is often  impractical, in part due to the cost of including expensive subject matter experts (SMEs) in the development process. In this paper, we describe a  method for “bootstrapping” model building to lower the cost of overall model development. The  models we are interested in here capture dynamic  phenomena related to international and subnational conflict. The method of acquiring these  models begins with event data drawn from news  reports about a conflict region, and infers model  characteristics particular to a conflict modeling  framework called the Power Structure Toolkit  (PSTK). We describe the toolkit and how it has  been used prior to this work. We then describe the  current problem of modeling conflict and the empirical data available to learn models, and extensions to the PSTK for model generation from this  data. We also describe a formative evaluation of  the system that compares the performance and  costs of models built entirely by an SME against  models built with an SME aided by the automated  model generation process. Early results indicate at  least equivalent prediction rates with significant  savings in model generation costs