15-day field study in an industrial C++ repository: after AI mass remediation, CI and review became the main bottlenecks, and directory-based batching with a per-change file cap restored throughput
In a 15-day exploratory single-case field study in a closed-source industrial C++ repository, an experienced developer used a command-line AI coding buddy to remediate widespread issues, triangulating Gerrit metadata with a developer diary and team chat through descriptive statistics and qualitative coding; AI-assisted remediation rapidly generated hundreds of commits touching thousands of lines, saturating build-on-commit CI and reviewer attention, naive per-file commits overloaded CI, and switching to directory-based batching with a cap on files per change restored throughput, yet still required explicit review solicitation, negotiation of acceptable commit granularity, and iterative follow-up to resolve build and static-analysis failures, showing that when mechanical editing becomes che