Issue Summary: The client finds the current interface for providing feedback (up/down voting and re-categorizing) on AI Triage results to be "cumbersome" and inefficient for high-volume reviewers.

Detailed Context: Rudy suggested that the interface should allow managers to simply state "what was wrong" rather than being forced to redo the entire triage process manually. The goal is to make the feedback loop faster so that leadership can provide more training data to the AI without it becoming a time-sink.