The rapid adoption of hybrid work by firms has led to a debate between managers and workers on the relative value of remote and in-person communication. Colocation between workers may be helpful for communication, aid with coordination, and affect the intensity of monitoring of workers by managers. Exploiting a hybrid work field experiment involving HR workers and using unique data related to the text of electronic communication between employees, this paper provides causal evidence of how colocation between employees affects internal communication within firms. A machine learning analysis of email content reveals that colocation is a substitute for horizontal, coordination-related communication, but---somewhat surprisingly---a complement to vertical, monitoring-related communication.