The VS Code team shares their journey of shifting from monthly to weekly releases driven by AI advancements, highlighting how AI-powered tools and workflows have accelerated development, improved quality, and automated issue management. They emphasize that this rapid cadence required fundamental changes in engineering processes, collaboration, and planning to maintain performance, stability, and team alignment in a fast-evolving landscape.
In this talk, Pierce and Josh from the VS Code team discuss their experience transitioning from monthly to weekly releases, driven largely by the rapid advancements and integration of AI technologies into their product. They highlight how AI adoption, particularly through agents, has significantly accelerated development and improved the product, but also introduced new challenges such as increased issue volume and the need for faster iteration cycles. This shift necessitated a fundamental evolution in their engineering processes, planning, and collaboration models to maintain product quality while moving at a faster pace.
Josh showcases several AI-powered tools and workflows that have transformed the developer inner loop at VS Code. One notable example is a tool that uses AI to validate UI changes by running component tests and generating screenshots directly within pull requests, greatly speeding up development and reducing manual testing overhead. Additionally, the team leverages AI agents for prototyping new features, enabling rapid experimentation and communication without the traditional overhead of lengthy specification documents. These innovations have helped the team iterate quickly and maintain high confidence in their changes.
The team also emphasizes the importance of maintaining performance and stability despite the accelerated release cadence. They have developed specialized AI-driven skills to monitor performance regressions and automatically detect and fix errors from telemetry data. This proactive approach to error detection and resolution, including automated triaging and duplicate issue identification, helps the relatively small VS Code team manage the large volume of issues generated both internally and by the community, ensuring the product remains reliable at scale.
Behind the scenes, the VS Code engineering system has been extensively enhanced to support these AI-driven workflows and the increased release frequency. The system integrates data from GitHub issues, telemetry, and scheduled jobs to automate triage, assign ownership, and even propose fixes for bugs. The team has built custom dashboards and tools, including a Chrome extension, to streamline issue management and maintain pipeline health. These improvements reduce manual effort and allow engineers to focus more on building features rather than managing operational overhead.
Finally, the team discusses how their collaboration and planning models have had to adapt to the fast-moving AI landscape. Traditional long-term planning gave way to more dynamic, short-lived workstreams focused on specific priorities, with frequent communication and daily stand-ups to stay aligned. This flexible approach helps the team respond quickly to new developments and shifting priorities but also introduces new challenges in coordination. Overall, the VS Code team’s experience underscores that successfully leveraging AI requires not only adopting new technologies but also fundamentally rethinking engineering processes and team collaboration.