Disputes and blind reviews

Your people can question what the AI decided in two ways: disputes and blind reviews. Managers and admins handle both under Reviews. Agents dispute from the call.

Disputing a score

1. An agent opens the call and chooses Dispute on a score.
2. They can add a note, and choose Send dispute. While it’s open, they can edit or withdraw it.
3. A manager or admin decides: Score stands or Agent was right.

What the agent sees

Once a manager decides, the agent sees the outcome on the call:

• Reviewed · score stands: the manager agreed with the AI. If something new comes up, the agent can dispute again.
• Reviewed · you were right: the manager agreed the AI got it wrong. The score stays, but it counts against the AI in AI agreement.

The manager can add a Note for the record. Managers and admins see it. The agent doesn’t.

Blind reviews

In a blind review, someone who manages a team scores calls picked for them each week. The AI’s score stays hidden until they submit, so it can’t influence them.

Each review in the Blind reviews tab says why the call was picked:

• A score on it was disputed
• The AI was unsure of a score
• Random pick

To review a call:

1. Choose Start review.
2. Score every skill from 1 to 5, or choose Not applicable when it doesn’t fit the call. Mark every check Pass, Missed or Not applicable.
3. Choose Submit review.

Scores are saved only when you submit, so leaving early discards them. A review that isn’t submitted in time expires, and the call may be picked for you again later.

After you submit, you see your scores and the AI’s side by side, with its reason for each. Your scores now count toward AI agreement.

While a review is open, the AI’s scores for that call stay hidden from you on the call, in call lists and in disputes, and you can’t decide a dispute on it. Team averages still include them.

AI agreement

AI agreement compares the AI’s scores with your people’s scores:

• Strong: the AI and your people mostly agree.
• Fair: they agree often, but not always.
• Needs calibration: they disagree too often, and the wording of the skill may need a review.

The tab shows the overall result, then each skill and check with its trend. A skill shows Gathering data until enough scores have been compared. A blind review agrees when your score is close to the AI’s, and a check agrees only when it matches. A dispute agrees when the score stood.

The tab also shows whether reviewers disagree more often when the AI was unsure.

Admins can review the wording in the scorecard. Managers can ask an admin to do it.