When a list runs against a Human in the Loop scorecard, any score that triggers a rule is held back and marked for a QA Team member. You review or complete it from the list page, then release it to the Agent as normal.
Find scores that need you
Open the list. The Scoring progress bar shows how many line items are processed and how many are Failed, Pending, Needs Review or Needs Completion. The Scoring Status column shows each line item's status.
Status | What it means | What you do |
Needs Review | AI scored every question, but the Overall Score Rule or an Answers that Flag for Review rule was triggered | Check the AI's answers, change any you disagree with, then release |
Needs Completion | AI skipped one or more excluded questions | Answer the excluded questions, then release |
The Result column shows the provisional score so far.
Review or complete a score
On the list page, select the line item (or Complete for a Needs Completion item). The scoring panel opens with the conversation on the left and the scorecard on the right.
Read the conversation. Use the search box to find specific words.
Work through the scorecard. Questions needing your attention are marked; tick Show only questions needing review to hide the rest. The panel header shows the provisional score and why the score was flagged.
Change or add answers and comments as needed.
Select Continue.
Release the score to the Agent
After Continue, the panel shows what will happen when you finish. Nothing is sent until you select Finish scoring, and you can go back to change any answer.
Under Visibility, choose whether to show the result on the Agent's dashboard.
Under Notify, choose whether to email the Agent, the Agent's supervisor, or both.
Under Recognise & coach, optionally give a kudos point or create a coaching session.
Select Finish scoring.
The final score then reaches the Agent in the same way as any other score.