Import your website
Your AI answers customer questions using your Knowledge Base. The fastest way to fill it is to import pages straight from your website. This page covers website imports; for PDFs, Word documents, and other files, see Upload documents.
Before you start
You’ll find the Knowledge Base in the admin app at app.jiviq.com under Settings → Knowledgebase.
- Only owners and admins can change the Knowledge Base. Members and viewers can see it but can’t import or remove content.
- Pages must be publicly visible. Anything behind a login can’t be imported.
If you don’t have a knowledge source yet, click Add knowledge source, give it a name (for example “Help docs” or “Website FAQ”), and pick the Workspace it belongs to. If you have one Workspace, it’s selected for you. Then open the source and use the Import pages card.
What you can import today
The Import pages card gives you three URL types. Pick one, paste a public URL, and click Import pages.
| Source type | What it imports | Use it when |
|---|---|---|
| sitemap.xml | Every page your sitemap lists, up to 1,000 pages per import | Your site has a sitemap (most do — see below) |
| llms.txt | The docs, blog, pricing, and general pages your llms.txt index lists | Your site publishes an llms.txt content index |
| Single page (no crawl) | One page, by its exact address | You want one specific page: a help article, your pricing page, an FAQ |
For most businesses, the sitemap is the right choice. A sitemap is a standard file most website platforms publish automatically. Sitemap index files work too.
Only want part of your site?
With the sitemap option you can add include patterns, one per line, for example:
/help/
/docs/
A page is imported only if its address contains at least one pattern. Leave the box blank to import everything the sitemap lists.
”Can I just add my home page link and you could crawl the whole website?”
Not today. Jiviq doesn’t crawl from a starting page yet — pasting your home page with Single page (no crawl) imports exactly that one page and nothing else.
The closest thing today is your sitemap:
- In a browser, try
yoursite.com/sitemap.xml. - If a page full of links or XML loads, you have one. Paste that address into the Sitemap URL field and import.
Most website builders (WordPress, Shopify, Squarespace, Wix, and others) generate a sitemap automatically, so this works for the large majority of sites. If yours has no sitemap and no llms.txt, import your most important pages one at a time with Single page (no crawl) — pricing, FAQ, services, and contact pages cover most customer questions.
What the import states mean
While an import runs, each page shows one of six states:
| State | What it means |
|---|---|
| Queued | Waiting its turn to import. |
| Importing | Being read right now. |
| Ready | Imported. Your AI can use this page now. |
| Already up to date | Unchanged since your last import. Not an error — there was nothing to do. |
| Finishing up | The AI service was briefly busy, so this page is waiting for the next automatic attempt. No action needed. |
| Couldn’t import | This page couldn’t come in. See the next section for why and what to do. |
“Finishing up” is progress, not failure. The import finishes itself automatically, and the panel shows when the next attempt will run. You can close this page, nothing will be lost. If you’d rather not wait, a Finish now button appears whenever pages are finishing up.
The summary line under the progress bar always shows three honest buckets: “X ready · Y finishing up · Z couldn’t import.” Pages that are finishing up are never counted as failed, and never counted as done until they truly are.
How long does it take?
An import usually takes a couple of minutes. Larger sites take longer. Everything runs in the background, so you can leave the page and come back.
If the AI service is busier than usual, some pages park in Finishing up and the panel tells you when the next attempt runs (for example, “Next attempt in about 12 minutes”). If an import is taking noticeably longer than usual, the panel says so and confirms it’s still going.
When a page couldn’t import
One page failing doesn’t affect the rest of your import. The two common reasons:
- “No readable content on this page.” The page loads its content with JavaScript, is empty, or duplicates a page already imported. If the page does have content you want, check that it’s publicly visible, then use Refresh this page on that row.
- “Couldn’t read this page — it may require a login.” Make sure the page loads for anyone, with no login or redirect, then refresh.
Rarely, a page shows “Couldn’t import this page right now” after the AI service stayed busy through several attempts. Try it again later — the rest of your import is unaffected.
Refresh after you change your website
Imported content is a copy taken at import time. When you edit your website, the Knowledge Base doesn’t update by itself — click Refresh from website at the top of the knowledge source page.
- Refresh reuses the source you saved (the same sitemap, llms.txt, or page URL), so it’s one click.
- Pages that haven’t changed come back as Already up to date, so a refresh is usually much faster than the first import.
- The panel shows Last refreshed with a timestamp, so you always know how fresh your AI’s knowledge is.
- While an import is running, refresh is paused until it finishes.
After the import
Every imported page appears in the Documents list on the knowledge source page — that list is what your AI actually knows. Pages marked Ready are used in answers immediately; there’s nothing to turn on. To understand how the AI picks what to say, see How the AI answers.
To remove a single page, use Remove on its row in Documents — your AI stops using it right away. To remove a whole knowledge source, use Delete knowledge source on the Knowledgebase page. That permanently removes the source and everything it contains, and your AI stops using that content until you add a replacement.