Import from URL
Crawl your existing docs or website into the knowledge base, choosing exactly which pages to bring in.
If you already have a help centre, documentation site, or blog, you can crawl that content into Kerabie's knowledge base instead of rewriting everything from scratch.
How it works
- Go to Knowledge Base → Crawl a website
- Enter the URL of your existing site
- Kerabie discovers pages on that domain (from on-page links, and from
sitemap.xmlwhen one exists) and shows you the list - Pick exactly which pages you want imported; nothing is crawled until you choose
- AI extracts and structures each selected page's content into draft articles
- Review and publish the ones you want
Discovery is instant and free. The actual crawl runs in the background (it can take a little while for several pages plus AI processing), so you'll see a progress state while it finishes.
What gets imported
| Content | Imported? |
|---|---|
| Page title | Yes → article title |
| Body text and headings | Yes → article content |
| Images | No: images aren't imported; add a cover photo to an article afterward if you want one |
| Navigation, footer, header | No: stripped automatically before AI sees the page |
Page limits
The number of pages you can select in one crawl run depends on your plan:
| Plan | Pages per crawl |
|---|---|
| Free | Not available |
| Basic | 20 |
| Pro | 100 |
| Enterprise | Unlimited |
This is a per-run cap; a page you don't select in one run can still be picked up in a later one. Crawling also counts against your plan's shared monthly AI-operations budget (import, crawl, and AI refine draw from the same pool).
After importing
All imported articles start as drafts. Review each one before publishing:
- Remove any leftover navigation text the AI didn't catch
- Update the category
- Add a cover photo if you want one; crawled pages never bring their own images
- Optionally run Refine with AI on an article to tighten the writing further
Re-crawling
Running a crawl again on pages you've already imported creates new draft articles rather than updating the existing ones; archive or delete the old version first if you want a clean replacement.