A legal agreement is a type of procedural document that describes the steps that parties must take to fulfill legal obligations. Following these steps requires human interpretation, which is often inefficient and error prone. For a Small to Medium Sized Business (SMB), this process is laborious. We conduct an exploratory design probe to understand how information in agreements is currently understood, processed, and taken action on by SMB employees working in small teams of non-domain experts. Through qualitative analysis and a text highlighting activity, we observe knowledge transfer workflows in small businesses and propose design principles for using AI-extracted information to create actionable documents that address gaps in efficiency, understanding, and agency.