Sentry recently launched its AI Code Review tool, which has already identified over 30,000 bugs within a month of use. For example, it caught a scheduling logic error in a side project by Staff Engineer Ryan Brooks, where two training phases could overlap. It also flagged a significant issue in the onboarding flow that could have prevented users from editing race details. These examples highlight the tool’s effectiveness in improving code quality. The tool's performance has also improved, with review times cut by about 50%. Sentry achieved this by switching to more efficient models for simpler tasks, setting limits on reasoning time, and conducting evaluations to ensure quality remained high. The structure of comments has been revamped to enhance clarity, providing a detailed analysis of issues, suggested fixes, and a new AI prompt for quick resolution. A feature called the Claude Skill automates the process further by connecting AI Code Review outputs directly to the codebase, allowing for quick patch generation without the need for manual copying and pasting. Sentry also hosted a workshop to help users understand the tool better, making it accessible for those interested in improving their coding workflows. Overall, the updates reflect a strong push towards faster, more reliable code reviews.