Adoption and Impact of Command-Line AI Coding Agents at Microsoft [pdf] (arxiv.org)

Microsoft's rollout of command-line AI coding agents, specifically Claude Code and GitHub Copilot CLI, early in 2026 has shown significant implications for software development practices. A study examining the adoption and impact of these tools among tens of thousands of engineers revealed that initial use spread predominantly through social networks, emphasizing the role of peer influence in technology adoption. Notably, the findings indicated that engineers who adopted these tools merged approximately 24% more pull requests than their non-adopting counterparts, suggesting a tangible improvement in productivity and coding efficiency sustained over four months. This research underscores the importance of understanding user dynamics when integrating new AI tools, especially given the substantial financial investments that organizations make in such technologies. Misjudging adoption rates or retention could lead to costly rollouts without substantial enhancement in engineering output. As CLI coding agents prove effective beyond novelty, organizations are encouraged to leverage visible peer usage as a central strategy in their deployment plans, ultimately paving the way for more informed and effective adoption of AI tools in software development.