Fourth Judge
When the AI, the BPO's QA and the client's QA disagree, ask the customer.
The job
Settles disputes over outcomes, like whether an issue was resolved, between the AI grader, the outsourcing partner's QA and the client's QA, by asking a sample of consenting customers. Their answers are the proposed contract tie-breaker and retune the grader. Compliance checks stay with humans.
The moment
The screen is set out like a courtroom with four judges: the AI grader, the outsourcing partner's QA team, the client's QA team and the customer. One call marked "resolved on first contact" has been argued over for three weeks.
Both QA leads lock in their verdicts. The supervisor taps Call, and an AI voice agent asks the customer, "Was your card problem solved?"
A minute later the customer answers that it works now and was sorted the same day. The point goes to the outsourcing partner.
Across 200 settled disputes, customers sided with the partner's QA 64% of the time, the client's QA 21%, and the AI grader 58%, rising to 71% after one retune. The app then drafts tie-breaker wording for the client's contract.
What it does
- Import grader verdicts and open disputes
- Lock a grader verdict
- Call the customer
- Propose an AI grader retune
- Send the tie-breaker clause for approval
What you see
Courtroom bench with four chairs; the fourth is a phone that rings live.
What it moves
How often each grader agrees with customers on each outcome question (with sample size and response rate), and the number of grading disputes closed each week.
Built for
- QA, training & knowledge
- CX leaders & VoC
- Frontline staff
