I’m currently evaluating the Agent Assist AI functionality for our contact center. During live calls, the agents see the knowledge base suggestions pop up correctly in their interactions panel.
However, when our QA team goes to evaluate those exact calls the next day using a customized evaluation form, the ‘Agent Assist Insight’ section of the QM UI is completely blank. We want evaluators to be able to see exactly which articles the AI recommended to the agent during the call to judge their adherence to the suggestions. Is there a specific retention policy or permissions issue preventing the live assist data from bridging over to the historical QM evaluation view?
I manage over 50 flows and integrate heavily with the knowledge base.
The issue here is usually related to the ‘Transcription’ requirements. Agent Assist relies on real-time transcription to generate suggestions. While the live suggestions are pushed to the agent’s browser via WebSockets, the historical record of those suggestions is tied to the saved transcript of the call. If your Recording Policy is set to record the audio, but ‘Voice Transcription’ is NOT enabled for that policy, the underlying text data that anchors the historical Agent Assist timeline is discarded after the call ends, leaving the QM view blank.
I’m just a supervisor, but we had a very similar issue when we rolled this out to our 30 agents.
Make sure the evaluators actually have the quality:evaluation:viewAgentAssist permission assigned to their role. Even if the transcription is saved and the data is there, the QM module hides the AI insights panel by default to prevent clutter unless the user explicitly has the rights to view the AI metadata. Check your QA roles!