Machine-learning models that automatically detect bugs, security vulnerabilities, and code quality issues during the pull-request review cycle.