The ability to faithfully model complex processes lies at the heart of experimental biology. Although a reductionist approach necessarily reduces this complexity, it is nevertheless required for untangling the contributions and interactions of the various system components. It has long been appreciated that cancer is a complex process that involves positive and negative interactions between tumor cells, normal host tissue, and the associated cells of the tumor microenvironment. However, accurate models for studying these complex interactions in vitro have remained elusive. We seek to generate models of mouse and human pancreatic cancer that are relevant to disease biology and useful for elucidating poorly understood facets of this deadly disease. The ability to model, manipulate, and predict the therapeutic response of an individual's disease outside their body represents the promise of precision medicine. Therefore, these models are patient-specific and allow the identification of new biomarkers and novel treatment modalities for rapid translation to the clinic. In this perspective we will discuss recent advances in modeling pancreatic cancer in vitro, the discoveries these models have enabled, and future challenges and opportunities awaiting further investigation.