TABLE OF CONTENTS

• Prerequisites
• Supported knowledge types 
  • 1. URLs
  • 2. Files
  • 3. Solution articles
  • 4. Custom Q&As
• Remove knowledge types
• Error handling & support

Your AI Agent’s response quality depends directly on the knowledge it is trained on. By configuring and curating knowledge in AI Agent Studio, you enable the agent to learn from structured and unstructured content such as URLs, files, solution articles, and custom Q&As.

A well-maintained knowledge base ensures:

• Higher response accuracy
• Reduced fallback messages
• Fewer unnecessary escalations
• Better customer experience across channels

This article explains the supported knowledge sources, how to add and manage them, how to disable or restore URLs when content becomes outdated, and best practices for maintaining high-quality AI responses.

Prerequisites

• Ensure you have Administrator access to AI Agent Studio.
• Ensure the AI Agent is already created.
• Ensure URLs are publicly accessible.
• Ensure files are not password-protected and meet size requirements.

Supported knowledge types 

The following knowledge types are supported:

• URLs
• Files
• Solution articles
• Custom Q&As

1. URLs

URLs allow the AI Agent to learn directly from publicly available web pages, such as help centers, product documentation, policy pages, or FAQ sections.

What the AI Agent learns from URLs

• Only static, crawlable text content
• Structured content such as headings, paragraphs, and lists

The AI Agent does not learn from:

• Videos
• Screenshots
• Animations
• Gated or authentication-protected pages
  
  

URL Limits

1. Go to AI Agent Studio > select the AI Agent. 
2. Click Build > Knowledge > Add URL.
3. In the Add URLs page, you can,
   • Select from previously added URLs, or,
   • Add a new URL not part of any of your AI Agent's knowledge base.
4. In the Add new URL pop-up, enter a name, add the URL, then select Add URL.

Notes: 
1. It can take up to 30 minutes for Freddy AI Agent to learn from a URL. You will receive an email once the learning is complete.
2. Ensure the page doesn't have pop-ups, CATPCHA or content blockers—these can interfere with the agent's ability to crawl. Check with your Dev/IT team to confirm permissions.

Manage URLs

As your website evolves, certain pages may become outdated, deprecated, or irrelevant. If outdated URLs remain in your AI Agent’s learning pool, they can negatively impact response accuracy.

Access URL Management

1. Go to AI Agent Studio > select the AI Agent.
2. Click on Build > Knowledge > URLs. 
3. You’ll see all configured URLs along with:
   • Learning status (Successful / Failed)
   • Last synced date
4. Click the ellipsis next to the configured URL to manage it.
   

Available Actions

• Show as references: Toggle this option to control whether the URL appears as a citation in AI responses.
  Enabled: The URL is displayed as a reference in answers.
  Disabled: The AI Agent continues learning from the URL, but it will not show it as a visible source.
• Resync and relearn: Use this option when website content has been updated. This forces the AI Agent to crawl the URL again and refresh its training data to reflect the latest content.

  Note: If you resync the website later, previously removed URLs will remain excluded from training unless manually restored.

• Download learning details: Download the processed learning data for auditing or review. This helps you verify what content the AI Agent has extracted from the page.
• Remove: Removes the selected URL from the AI Agent’s learning pool. When you remove a URL:
  • All pages within the same URL path are excluded from training.
  • The AI Agent will no longer reference that content in responses.
  • Other active URLs remain unaffected.This allows you to disable outdated or irrelevant content without affecting the rest of your knowledge types.

Review URL performance

Click Manage URLs to review learning performance, refresh content, and control how each URL contributes to the AI Agent’s responses. The Manage view of a particular URL displays:

1. Learning status (Successful / Failed counts)
2. Last synced timestamp
3. URL content preview
4. URL-specific actions -Remove from agent knowledgeThis provides visibility into how effectively the AI Agent has processed the website content.

Restore removed URLs

If a removed URL is updated and becomes relevant again, you can restore it by following the steps below,

1. Navigate to the URLs tab > select Manage URLs from the URLs list.

2. Click on Removed URLs.

3. In the Removed URLs overlay, select the required URL(s) and click Restore.
   

4. Confirm the action.

   Restored URLs are reintegrated into the AI Agent’s training data and retain their prior learning status.

   Note: Restored URLs retain their previous learning status and are reintegrated into the AI Agent’s training data.

Regularly reviewing URL learning status, syncing updated content, and removing outdated pages ensures your AI Agent delivers accurate, current, and trustworthy responses.

2. Files

• Formats supported: .pdf, .docx, .txt
• Size limit: 35 MB per file
• Security: Password-protected files are not supported
  

File Upload Limits

1. Go to AI Agent Studio > select the AI Agent. 
2. Click Build > Knowledge > Add files.
3. In the Add files page, you can,
   • Select from previously added files, or,
   • Upload a new file not part of any of your AI Agent's knowledge base.
4. Click the ellipsis next to the file name under the Available files list to manage it.

Note: To prevent a file from being shown as a citation in the answer, disable the Show as references toggle. This will ensure that the file is not displayed as a reference in the response, although the AI agent will still learn from its contents and use it to give answers.

3. Solution articles

Solution articles allow your AI Agent to learn directly from your existing Knowledge Base (KB). This is ideal if you already maintain a well-organized support portal and want the AI Agent to respond consistently with your documented processes and policies.

Toggle Learn from solution articles to enable the feature and provide the following details:

• Portal selection: Choose the knowledge base the AI Agent should learn from (if multiple portals are available).
• Select your AI's learning scope: Choose how broadly the AI Agent should learn:
  • All categories – Learns from all articles in the selected knowledge base.
  • Selected categories – Learns only from specific folders or topics (use Manage articles to choose).
    Note: Only articles visible to All users are used for AI learning.
• Display Article References: Enable the toggle to show article links in AI replies.

Note: Category selection is supported only in old Freshdesk Omnichannel accounts, not in Freshdesk Omni (formerly Customer Service Suite-CSS) or Freshchat standalone accounts.

4. Custom Q&As

Add targeted content that may not be available in existing knowledge sources.

Example:

Q: What is the charging time for the battery of the Z1 Zola electric scooter?

A: The Z1 Zola electric scooter takes 4.5 hours to fully charge with a home charger. The Hypercharger gives up to 60 km range in 10 minutes.

Click New Q&A to include new content. You can edit or delete them anytime.

Remove knowledge types

To permanently delete or remove a file or URL:

1. Navigate to Admin Settings > Conversation Resources > Files / URLs.
2. Locate the specific file or URL you wish to remove.
3. Click the three-dot menu next to it and select Delete.
   

Note: Removing a file or URL from this section permanently deletes it from the AI Agent’s learning pool. This means the bot will no longer refer to this source for any responses. Be cautious while deleting, as this action cannot be undone.

Error handling & support

• Issue with file/URL not loading: Ensure there are no blockers or authentication on the page.
• Pop-ups interfering with learning: Reach out to support@freshdesk.com for resolution.
• Need to extend file or URL limits: Contact support@freshdesk.com - requests are evaluated based on your business use case.

Tip: For best practices and guidelines on writing knowledge sources, see Write AI-optimized knowledge content.