Automating migration reporting

A content migration produces plenty of technical output. The people following its progress need a clear view of what has moved, what remains and where attention is needed. The people investigating problems need the underlying detail.

I used AI-assisted development to build a Python reporting tool around both needs. I shaped the reporting questions, checked how source and destination counts should be compared, and kept challenging the output when it read like a developer log.

The tool combined content counts with import feedback and retained measurements from earlier runs. Scheduled delivery produced separate business and technical workbooks. The overview showed progress and completion estimates, while the detailed report supported investigation of errors and individual records.

The useful distinction was between the report someone needs to understand the migration and the evidence someone needs to investigate it. Both came from the same reporting process.

AI helped with implementation and iteration. The decisions about what to count, which discrepancies mattered and how to present the result came from examining the actual migration and discussing the reporting needs.

The tool generated reports and ran on a schedule. As its history grew, a later workbook repair addressed memory use while retaining earlier measurements. That repair was checked locally. Its next scheduled run was not verified in the available record.