In this study, we use word and document embedding techniques from natural language processing to contextualize policymaking in U.S. federal agencies. We specifically look at a range of agencies responsible for economic regulation and train a model that maps the language of rulemaking to vector spaces. Specifically, we are interested in contextualizing the way different agencies talk about regulation, and juxtaposing agency text with citizen comments. Our mapping will help uncover the ways that different agencies approach policymaking and regulation, how citizens’ views on regulation differ from agencies’, and how much final rules reflect an interplay between agencies and commenters. These results will enrich the current literature around administrative rulemaking and regulation by providing a rich, nuanced understanding of the policymaking process.