Stanford CS146S Instructor Mihail Eric on How AI Coding Agents Are Reshaping Engineering Curriculums and Software Factories

Stanford instructor and Monaco AI lead Mihail Eric shares practical perspectives on teaching AI-native engineering in the era of advanced coding agents. Facing a reality where students can finish last year's assignments using a single prompt, Stanford's CS146S course has redesigned its curriculum by bringing in open-source maintainers. Students must now comprehend unfamiliar repositories, propose patches, and defend their code to actual maintainers, shifting the educational bottleneck from writing code to understanding and evaluating a diff. At Monaco, Eric's team builds bespoke agent workspaces, integrating on-call agents with tools like Datadog and GitHub. He argues that the future of open-source repositories will transform them into agent-run software factories. In this shifting landscape, the role of engineers is pivoting from manual coding to supervising background agents, managing context engineering for large codebases, and navigating automated code reviews.