Kustomer's AI QA scores each conversation an AI agent finishes against the admin's scorecard. For example, a scorecard can require confirming a resolution timeline. A low score comes with reasoning that shows where the agent missed.
A workflow flags the low score to an admin right away. The admin reads the reasoning and updates the agent's instructions. Human reps are scored against the same scorecard.
When an AI agent hands a conversation to a person midway, AI QA scores the whole interaction as one conversation. New customers can book a demo to see it on their own conversations. Existing customers should contact their Customer Success Manager for rollout timing.