Zoom Virtual Agent admins can coach queries to improve model accuracy by collecting chat messages from customer interaction. The queries are added to the Coach page, and admins can either add them as training phrases, move them to another intent, or mark them as junk if they're irrelevant.

This article covers:

• How to add queries as training phrases for intent
• How to move queries to another intent
• How to filter out queries

Prerequisites for coaching queries

• Account owner or admin privileges; or relevant role/privilege
• Basic, Pro, Business, Education, or Enterprise account
• Zoom Virtual Agent license
• Bot flow created

How to add queries as training phrases for intent

1. Sign in to the Zoom web portal.
2. In the navigation menu, click AI Management then Intent Management.
3. Click an intent group.
4. Click the Coach tab.
   You will see a list of queries that the model has determined will be most helpful in improving the model accuracy.
5. Enter a keyword in the search bar or select a query from the list. You can also narrow down the list by specifying start and end dates.
6. Click Add to Training Phrases.
   The selected queries will be added to the Training Phrases tab.

Note: Once the selected queries have been added as training phrases, you can now train and publish the intent group.

How to move queries to another intent

1. Sign in to the Zoom web portal.
2. In the navigation menu, click AI Management then Intent Management.
3. Click an intent group.
4. Click the Coach tab.
   You will see a list of queries that the model has determined will be most helpful in improving the model accuracy.
5. Enter a keyword in the search bar or select one or multiple queries from the list.
6. Click Move to another intent.
   A pop-up window will appear.
7. Select an intent name from the drop-down.
8. Click Save.

How to filter out queries

1. Sign in to the Zoom web portal.
2. In the navigation menu, click AI Management then Intent Management.
3. Click an intent group.
4. Click the Coach tab.
   You will see a list of queries that the model has determined will be most helpful in improving the model accuracy.
5. Enter a keyword in the search bar or select one or multiple queries from the list.
6. Click Mark as Junk.