Agentic Automation and the Reorganization of Human Work in Small Open-Source Software Teams
Type: article
Abstract: AI coding agents mark a shift from workflow automation to agentic automation, in which AI systems participate as semi-autonomous collaborators in software development. While prior research has focused on productivity and efficiency, less is known about how agentic automation reshapes human work within teams. This study examines changes in team structure and work allocation following AI-agent adoption in open-source software (OSS) projects. Using a panel of GitHub repositories and a propensity-score-matched difference-in-differences design, we analyze the short-run consequences of Claude Code usage. We find no statistically detectable reduction in active human contributors. Adoption is associated with a broader functional scope of human work, with a more pronounced increase in development-heavy projects. These findings indicate that the early organizational consequences of agentic automation are reflected in the reallocation and broadening of human work, highlighting augmentation and work reorganization as important dimensions of AI-enabled software development.
Authors: Liu, Xinhui, Kang, Lele, Sia, Choon-Ling
Institutions: Nanjing University, City University of Hong Kong
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7222988873