Your agents are already tagging and you want to build a tag based on the initial tagging system you defined. Replicating those tags in Lang.ai can help save your agents a countless amount of time.
Here is the best process for replicating a Zendesk tag in Lang.ai:
Inside a project, add your tags in setup->metadata by selecting the field.
For example, in Zendesk, select the field tags and add a name to it.
Inside a project, click on Search and you'll see the ability to filter by those tags on the top.
Choose the tag that you want to replicate in that drop-down menu and filter by it.
On the left-hand part of the screen, you'll see the top Concepts table. These top Concepts are dynamic and they change based on the search. As you filter by tag, the platform will outline the top Concepts that are related to that tag.
Navigate through the table to find related Concepts with their own specific concepts (if you click on the arrows on the left, you can navigate through the hierarchy).
Once you see a Concept that you want to analyze, you can filter it by clicking on the number of comments on the right of the table. You can use those search results to create a new tag.
Once you've created a new tag, you can repeat this process with all of the Lang Concepts you want to add inside the tag.
To increase the coverage, we recommend adding the filter "documents with no tags" together with the initial metadata tag filter. This simple action will filter your tagged tickets to show you the ones that are not inside the Lang tag you're building. This will be really helpful in terms of understanding what new concepts need to be added to the tag.
The goal when replicating a tag should be that when you filter by the Zendesk tag and the Lang.ai tag, the results should be similar in the amount of documents/tickets. When you filter the Lang.ai tag with the "documents with no tags" filter, the results should be at almost 0 documents (this indicates that you've made an exact replica of the Zendesk tag).
*Keep in mind the Concepts on the left don’t represent complex logical rules, so you may need to combine them. Click here to read more.