Abstract: Techniques for identifying actual communications and data flow between users of an organization exchanged via a communication platform and providing a recommended action for a user to perform based on the actual communications are described herein. The communication platform receives messages exchanged between users of the organization and generates an interaction graph representative of actual interactions (e.g., communications, data flow, etc.) between the users. The interaction graph can be based on a frequency of messages, message content (e.g., message data, attached documents, links, applications, etc.), context data (e.g., personal, business, collaborative (e.g., associated with a project, etc.), etc.), user data associated with the users (e.g., position, title, team identifiers, etc.).