Knowledge sources
Websites
Import pages from your public website, one link at a time or from a sitemap, and keep them fresh automatically.
If your help centre, FAQ or product pages are already online, you don't need to copy them anywhere. Add a page's link, or import your whole site from its sitemap, and OrbitBase reads each page's text and trains your agent on it. Pages are re-read every 7 days, so your agent keeps up as your site changes.
Before you start
- The pages must be public. OrbitBase can't read pages behind a login, a paywall or a private network.
- The page's text must be in the page itself. Content that only appears after the page runs scripts in a browser — some single-page apps and embedded widgets — may not be picked up. If a page trains with little or no text, add that content as free text instead.
- OrbitBase respects your site's
robots.txt. If it blocks crawlers from a page, that page can't be imported.
Add a single page
Open Websites
In the sidebar, click Sources, then Websites.
Choose Individual links
Under Add pages, make sure the Individual links tab is selected.
Enter the page's address
In Page URL, paste the full address, such as
https://example.com/help/refunds. Add one page at a time.Add it
Click Add page. You'll see "Page added — training has started." The page appears in the list and changes to Trained when your agent can use it.
Import a sitemap
A sitemap lists every page on your site, so it's the fastest way to import a help centre or a large FAQ.
Choose Sitemap
Under Add pages, select the Sitemap tab.
Enter the sitemap's address
In Sitemap URL, paste its address — usually
https://yoursite.com/sitemap.xml. Sitemaps that link to other sitemaps work too.Import it
Click Import sitemap. OrbitBase shows Reading sitemap… while it works, then tells you how many pages were added. Training starts straight away.
How sitemap imports work
- Up to 50 new pages are added at a time. If your sitemap lists more, the rest are added automatically, up to 50 a day, until every page is in.
- Pages you already have are skipped, so you can re-import the same sitemap to pick up new pages without creating duplicates.
- Pages you've deleted stay deleted. A sitemap import won't bring back a page you removed. To restore one, add it with Individual links.
A sitemap usually includes pages your agent doesn't need, such as blog archives, legal pages or log-in screens. After importing, delete the pages that aren't useful for answering customers. They won't be re-added.
Keeping pages up to date
Every website page is re-read every 7 days. If a page has changed, your agent is retrained on the new text; if it hasn't, nothing changes.
To pick up a change straight away — after updating your returns policy, for example — open the page's row menu and click Refresh from the site. You can also click Train all to retrain every page that's waiting or failed.
To check a page's live version, open its row menu and choose to open it in a new tab.
Delete pages
Owners and admins can delete pages. Open a page's row menu and click Delete, or select several pages and click Delete. Your agent stops using them straight away, and they won't be re-added from your sitemap.
Allowing OrbitBase in robots.txt
OrbitBase identifies itself as orbitbase-bot. If your robots.txt blocks it, or blocks all crawlers, from pages you want to import, add a rule that allows it:
User-agent: orbitbase-bot
Allow: /help/OrbitBase also honours a Crawl-delay in your robots.txt and waits between requests, so an import never overloads your site.
Troubleshooting
"That site's robots.txt asks crawlers not to read this page."
Your site's robots.txt blocks the page. Import a page it allows, or update the file on your site as shown above.
"That page is already in your knowledge base."
You've already added this page. Find it in the list and use Refresh from the site to update it.
"No readable text found on that page."
The page loaded but had no text OrbitBase could use — often because its content is loaded by scripts, or the page is mostly images. Add the content as free text instead.
"That URL doesn't point to a web page." or "That URL doesn't point to a sitemap."
Check the address. In Individual links, use a normal web page. In Sitemap, use the sitemap file itself, which usually ends in .xml.
"That page took too long to respond." or "Couldn't reach that page."
The site was slow or unreachable. Check the address opens in your browser, then try again.
"Every page in that sitemap is already in your knowledge base."
There's nothing new to import. Your existing pages are still refreshed every 7 days.