What Web URLs and Links Do
Web URLs and Links allow Kyrios to crawl website content and add selected pages to a Conversation AI Knowledge Base. You can crawl:
One exact page
All URLs in a domain
All URLs with a shared path
You can train up to 4,000 web URLs in a single Knowledge Base.
Key Benefits
Web URL training helps Conversation AI use approved website content during customer conversations. Key benefits include:
Add website content without manually copying answers.
Train Conversation AI on approved customer-facing pages.
Choose exact, domain, or path-based crawling.
Review crawled URLs before training.
Refresh or delete trained URLs when website content changes.
URL Crawling Modes
Crawling modes control how broadly Kyrios searches for website pages before adding them to the Knowledge Base. Choosing the right mode helps prevent irrelevant content from being trained.
Exact URL
The exact URL trains one specific webpage. Use the exact URL when you only want Conversation AI to use one page. Use Exact URL for:
Pricing pages
FAQ pages
Service pages
Refund or cancellation policies
Single landing pages
Example: If you enter https://example.com/services, Only that exact page is crawled and trained.
All URLs in This Domain
All URLs in this domain crawl the website domain and return available URLs for review. Use this option when most pages on the website contain useful business information. After Kyrios fetches the URLs, review the list carefully before selecting pages to train.
All URLs with This Path
All URLs with this path crawl pages that share a specific URL path. Use this option when useful content is grouped under one section, such as:
/services/blog/help/support/locations
Example: If you enter https://example.com/services, Kyrios can return pages such as:
https://example.com/services/seohttps://example.com/services/web-designhttps://example.com/services/consulting
New Pages and Existing Pages
Fetched URL results may be grouped as New Pages or Existing Pages. New Pages are URLs found during the crawl that are not already part of the Knowledge Base. Existing Pages are URLs already included in the Knowledge Base. Select New Pages to add fresh content. Select Existing Pages to refresh content that was already trained.
Best Practices for Training URLs
Train only on clean, relevant, customer-facing website content. Recommended best practices:
Train accurate and current pages.
Avoid unrelated pages.
Avoid duplicate or low-value pages.
Review crawled URLs before training.
Use the exact URL for precise training.
Use Path crawling when only one website section is relevant.
Refresh or delete outdated links.
Wait for training to finish before testing the bot.
How To Set Up Web URLs and Links
Follow these steps to train Conversation AI using website URLs.
Step 1: Open Bot Training
Go to AI Agents.
Open Conversation AI.
Select Bot Training.
Step 2: Enter the URL and Choose a Crawl Mode
In the Web Crawler section, enter the full URL, including https://.
Choose one crawl mode:
Exact URL
All URLs in this Domain
All URLs with this Path
Step 3: Fetch Website URLs
Click Get Data. Kyrios will fetch the available URLs based on the selected crawl mode.
Step 4: Review and Train URLs
After Kyrios fetches URLs:
Review the fetched pages.
Select the URLs that should be used for training.
Click Train Bot.
Wait for the URLs to appear in the Uploaded Links table.
Confirm each selected URL has completed training.
URL Training Statuses
Training statuses show whether a URL is ready, processing, or needs attention.
Common statuses include:
Getting Data: Kyrios is processing or refreshing the URL.
Trained: The URL was successfully trained.
Failed: Kyrios could not train the URL.
If a URL fails, confirm the page is accessible and useful. Then refresh the URL or delete it if it is no longer needed.
Common Issues and Fixes
The bot is giving answers from unrelated pages
Review the trained URLs and remove pages that are not relevant to customer conversations.
A URL failed to train
Check whether the page is accessible, public, and still needed. Then refresh the URL or retry training.
Website changes are not reflected in bot answers
Refresh the trained URL from the Uploaded Links table so Kyrios can retrain the latest page content.
Too many URLs were fetched
Use Exact URL or All URLs with this Path for more controlled crawling.
FAQ
Should I train my entire website?
Only train your entire domain if most pages are useful for customer conversations. For better accuracy, select the most relevant pages and avoid unrelated or outdated content.
What is the difference between Exact URL, Domain, and Path crawling?
The exact URL trains one specific page. Domain crawling finds pages across the full website domain. Path crawling finds pages that share a specific URL path, such as /services /help.
Can I refresh a trained URL after my website changes?
Yes. Use the refresh action in the Uploaded Links table to retrain the URL with the latest page content.
Can I delete a trained URL?
Yes. Deleting a URL removes that page’s trained content from the Knowledge Base.
Why do I see Existing Pages after fetching URLs?
Existing Pages are URLs already included in the Knowledge Base. Selecting them refreshes their trained content instead of adding them as new pages.
What should I do if a URL fails to train?
Check whether the page is accessible and still needed. Then try refreshing it. If the URL is not useful or cannot be processed, delete it from the Uploaded Links table.
How many web URLs can I train in one Knowledge Base?
You can train up to 4,000 web URLs in a single Knowledge Base.
Can I use Web URLs and Links with other Knowledge Base sources?
Yes. Web URLs are one type of Knowledge Base source. Kyrios also supports other Knowledge Base source types for Conversation AI.