Log predicted email classifications for data scientists before standup

Your predicted email records land in the warehouse uncaptured, causing inconsistent labels and delayed model feedback. Data scientists receive a cleaned, timestamped log to prioritize rechecks before the next analysis run.