Reviews Overview
A Review is when a user evaluates the AI Intent prediction of a specific message within a Journey. This gives users the ability to supervise the automation processes ("Human in the Loop"), to ensure that there a specific use case has 100% confidence prior to finishing a resolution path. Journeys do not require the review step and can be completely automated or have the Review step added at any point within a Journey Path.
Users might use the Review primarily as a methodology to make sure that risky or high priority cases are handled with a definitive intent classification. Any cases where a there might be significant financial loss or customers who have to be handled with care would be good candidates for the review process. Reviews require an agent or user complete an action, but are built to allow for quick and efficient handling that will still significantly reduce the time an agent has to spend on a message.
The Review process gives the agent key pieces of information shows the agent the relevant information needed to resolve the message:
Customer Message
Supplemental data (user ID, case type, users history and status, phone number, etc.)
Predicted Intent
Intent Confidence
Proposed Changes (list of changes that will happen when the review is approved)
Response Message (a response message is prepopulated by the system)
Reviews offer value by providing the ability to monitor the automation, use enrichment data to expedite manual resolutions, and refine the AI model when an intent is incorrect.
For more Review details, watch the video below:
Last updated
