Dataset Resources

Owned by SUPPORT
Last updated: Oct 15, 2024
2 min read

You can improve AI responses drafted by AI Assist and AI Agent by identifying various knowledge bases and files, that can help AI Assist/AI Agent work better. With sufficient material, AI Assist can provide apt responses to customer queries.

Create a Dataset

Step-by-Step Instructions

  1. Go to your system CloudCX installation domain, and log in with the admin account.

  2. From the left navigation menu, go to Automation > Agent Assist > Dataset Resources.

  3. Click +Create to create a dataset.

Note: Both AI Assist and AI Agent rely on existing dataset reources; any deleted dataset resource are not used to answer queries.

1-20240927-094500.PNG

  1. Input a name and description for the dataset, and click Create.

2-20240927-094903.PNG

  1. In the Edit Dataset page, you can do the following to add sources to the dataset. Successfully created, these datasets can be used by AI Assist/AI Agent.

Internal Knowledge Base:

  • Turn on the toggle to enable Al to learn from the support content in your knowledge base.

  • Select one or more knowledge base resources, then articles from your knowledge bases will be used in this dataset.

Click here, and learn how to manage your knowledge base.

5.1 Knowledge Base

External URL:

  • Turn on the toggle to enable Al to learn from imported content from public URLs, such as websites.

  • Click Add an External Content.

  • Input any public URL (such as pages on your website, pricing information, blog posts, etc.)

  • (Optional) You can click image-20240927-101110.png to unfold the input area, input a list of special URLs.

  • Click Add to add the external content.

External Files:

  • Turn on the toggle to enable upload files, and we'll fetch all the text data inside.

  • Click +Add a File to select a local file to upload.

Inbox Suggestions:

  • Turn on the toggle to enable Al to learn from conversations handled by your agents in the Inbox.

  • (Optional) Click View&Manage switch to Automation > Agent Assist > Inbox Suggestions to check in used suggestions. About Inbox Suggestions, you can find more details about Dataset Resources | Inbox Suggestions.

Manage Datasets

  • Edit: Select a dataset, click image-20240929-020354.png button and click “Edit” icon to open the Edit Dataset page.

  • Delete: Select a dataset, click image-20240927-091440.pngand click “Delete” icon to delete the dataset.

If the dataset has been used in AI Assist or AI Agent, you need to disassociate the belonging relationship of the dataset, then you can delete the dataset.

View & Manage Inbox Suggestions

Your Support team's conversations contain a wealth of knowledge with some of the most reliable and up-to-date information. With conversation content enabled, agents can suggest content to be used again by AI, directly from the Agent Panel.

Step-by-Step Instructions

  1. Go to your system CloudCX installation domain, and log in with the admin account.

  2. From the left navigation menu, go to Automation > Agent Assist > Inbox Suggestions.

  3. In the Inbox Suggestions page, you can do the following:

  • Click Settings to open the Settings page, you can do the following in the Settings page:

    • Turn on the toggle to enable collecting conversation content from the Inbox.

    • Choose which teammates are able to recommend conversation content. For example, you may only want to enable this for senior Support reps with more experience. Use the dropdown to select which teammates can recommend their conversation content, or allow Any teammate to recommend content.

    • Choose whether conversation content immediately available to Al or they need to be approved first. Once conversation content is recommended via the Help Desk, it will automatically be added as Snippets on the Content page. You'll need to decide if these Snippets are: Immediately available to AI Recommended, or Set them as 'pending' in the 'Content suggestions' page for review.

  • (Optional) If you select Set them as 'pending' in the 'Content suggestions' page for review in the Settings page, you can click image-20240929-071656.png button to review suggestions to open the first pending review suggestion on the list in a modal. This allows you to preview the content, edit the suggestion, skip to the next pending review suggestion, add as a suggestion, or reject the suggestion. After approving the suggestion, the status of the suggestion will be switched from Pending to Available.

  • In the suggestion list, you can do the following:

    • Edit: Select a suggestion and click Edit icon to open the edit page.

    • Delete: Select a suggestion and click the Delete icon to delete the suggestion. Also, you can select more than one suggestion from the left checking box and click Delete button to delete selected suggestions.