Flag product data messages for data team before model runs

Your product data messages marked with a done reaction sit unlogged, causing gaps in downstream models. Data scientists receive flagged, validated records so training and reports stay accurate before model runs.