If you’d like to train Fin with website content, you can do this by syncing the public URL for that site.

Go to Train > Content and then select Website sync below the "Add content" section.

Now enter the public URL of your website (top-level domain) and click Next.

This will fetch all of the pages from the website URL you provide and will read from all the sub domain pages.

Once you input your URL, we will check that it's valid and accessible. Then you'll need to review the pages to sync. All sub-pages linked in each selected section will be synced. Select only relevant up-to-date content.

Select the Advanced settings dropdown to configure additional URLs, exclude URLs, CSS selectors to exclude, and more.

Website structures can vary. To make sure that we sync your most relevant content, we recommend you add additional URLs for those specific subpages.

For example, if you input https://myhelpcenter.com/help as the primary URL above, you might also want to add the specific URL like https://myhelpcenter.com/help/index.html

To exclude certain pages you don’t want to sync content from, you can add a list of URL globs.

A glob is a string of literal and/or wildcard characters used to match file paths or URLs. Globbing is the act of locating files on a filesystem using one or more globs. Using URL globs also helps to get a range of URLs that are mostly the same, with only a small portion of it changing between the requests.

For example, this URL glob https://{store,docs}.example.com/** lets the crawler access all URLs starting with https://store.example.com/ or https://docs.example.com/ and https://example.com/**/*\?*foo=*

To exclude certain page elements, you can use CSS selectors of those specific sections or elements you want to exclude.

This is useful to skip irrelevant page content. The value must be a valid CSS selector as accepted by the document.querySelectorAll() function. By default, we already remove common navigation elements, headers, footers, modals, scripts, and inline images.

This allows for DOM elements identified by the CSS selector, to be clicked during the web sync process.

This is useful for expanding collapsed sections, in order to capture their text content. The value must be a valid CSS selector as accepted by the document.querySelectorAll() function.

Examples are "[aria-expanded=\"false\"]", #expand_section

Complex conditions can be also described with a CSS selector. In CSS, chaining the selectors without spaces creates an AND-like condition, for example .button.blue.small will match only elements with all three classes.

Using comma (,) as a separator works like OR, for example .button, .blue, h1 targets all elements with class button, or class blue, or first-level headings.

To target content that may have a delay in appearing on the page, you can add a CSS selector that will make the web scraper wait before scraping content.

This is useful for pages for which the default content load recognition by idle network fails. Setting this option completely disables the default behavior, and the page will be processed only if the element specified by this selector appears.

Enable XML Sitemap for a more robust websync on sitemap-supported websites, allowing access to pages that might not be reachable from the initial URLs.

If this option is enabled, the web scraper will look for Sitemaps at the domains of the provided source URL and enqueue matching URLs similarly as the links found on crawled pages. You can also reference a sitemap.xml file directly by adding it as another Start URL e.g. https://www.example.com/sitemap.xml.

You can select a proxy for crawler to use if you require your website to be crawled with a proxy set to a specific region or country.

We currently support the following proxies:

Choose how thoroughly the crawler renders each page while syncing this website.

When enabled, the web scraper ignores canonical link tags (rel="canonical") and treats each URL as a distinct page. This option is disabled by default.

This is useful when a site points many pages at one canonical URL and content is being missed as a result.

There are two more toggles under Advanced settings. Both are off by default.

Respect robots.txt: makes the crawler obey the disallow rules in your site's robots.txt. Turn it on to keep syncing consistent with what you allow other crawlers to reach. If robots.txt disallows paths that hold content you want in Fin, this will reduce what gets synced.

Exclude query parameters: skips any URL with a query string (?). Turn it on when your site appends tracking, session, or sort parameters that produce lots of near-duplicate pages. Leave it off if real content lives behind a query string, like /docs?page=intro — those pages would be skipped too.

The Target step lets you set a default audience for all synced pages and create URL-based rules to automatically assign specific audiences based on URL patterns — no manual tagging required.

First decide whether content from this source is enabled for Fin AI Agent and/or Copilot.

Then you can either set a default Fin audience to apply to all pages synced from this source (if no default is set, content defaults to Everyone), or create rules to assign specific audiences based on URL patterns.

For example: If URL contains /uk, assign the UK audience. Rules are evaluated during sync, so Fin and Copilot always serve the right content to the right audience.

Each rule supports three URL comparators:

Finally, review your sync settings then go ahead and click Sync website to start syncing your website content with Intercom.

Once the sync is complete, you’ll receive an email notification and the website will appear as a synced source in Train > Content under the "Content sources" section.

If you click into a website source, you can preview and manage the individual pages that were synced from the public URL.

When you view a website page, you’ll find a "Details" panel on the right which contains:

To make a website source available to Fin Agent or Fin Copilot, go to Train > Content and click on the website source under the "Content sources" section, then open the relevant webpage you've synced.

From the "Details" panel, scroll down to “Fin” and toggle on:

You can schedule when synced website pages turn on or off for Fin and Copilot, at a future date, time, and timezone. Set a start date only, or set both a start and end date to create a time-limited availability window. Scheduling applies to individual pages or in bulk.

You can assign audiences to synced content automatically using URL-based rules during web sync creation, or manually on a per-page basis. First, you'll need to create and define the audience you want to target.

Then go to Train > Content and click on the website source under the "Content sources" section, then open the relevant webpage you've synced.

From the "Details" panel, scroll down to “Fin” and use the audience dropdown to select one of your pre-defined audiences.

You don't need to recreate a sync to add audience targeting. Go to Train > Content, select the source, click the settings dropdown in the top right, and select Open settings. Navigate to the Target step to add or edit URL-based rules.

If you’d like to re-sync or remove a public URL as a source, go to Train > Content, and click on the website source under the "Content sources" section, then open the Settings dropdown in the top right.

Here, you can select whether to Re-sync or Remove this source.

You can view a list of past website syncs to see when they were last run, which pages were found, and any failed pages. Go to Train > Content, and click on the website source under the "Content sources" section, then select View sync history.

Each row in the table represents a past or active run, and you can filter the runs by status (started, success, failed).

It includes the following information:

If a sync has failed, you can hover over the status to see a detailed explanation for why.

A new source syncs on schedule for its first 90 days regardless. After that, it keeps syncing automatically if either is true:

Otherwise, scheduled syncs pause.

When paused: pages stay in your Knowledge Hub and stay available to Fin and Copilot (nothing is deleted). Only the re-crawl stops, so edits on your site won't be picked up.

To restore it: The Sync now button refreshes the content once but doesn't restart the schedule. Automatic syncing resumes on its own once Fin uses the source again.

Very large sites, and sites needing full JavaScript rendering, cost far more to crawl. When a source turns out to be resource-heavy, we move it from a weekly to an every 14 days schedule.

This happens automatically, per source. To keep a big site on a faster schedule, narrow the crawl by using URL exclusions, or add specific subpages instead of a whole domain.

When importing website content to enable Fin, you need to enter the public URL. This will search for all pages nested under that URL and sync them for Fin AI Agent to use.

If the importer didn't return the number of pages you expected, there are a few reasons...

The website sync works by going to the URL you provide and then searching for all pages nested under that URL. These pages must have the same URL pattern as the URL you provide.

For example, if the top level domain is https://myhelpcenter.com/home, then all pages you want to import must include /home prefix in the URL e.g. https://myhelpcenter.com/home/article. If they do not, remove the prefix and use the most basic URL stem e.g. https://myhelpcenter.com, then try the import again.

If the content you want to use is behind a login, Fin won't be able to access or import it.

Website sync has two limits:

A sync can also fail if a single page contains a very large amount of content. You'll be notified by email whenever a sync fails.

Fin's website sync (used to add public URLs for Fin AI Agent and Copilot) does not use a dedicated, custom user-agent string at this time.

If your site has strict anti-crawling protection, you can use a static proxy to allow-list a fixed set of IP addresses and ensure your content gets ingested. Select a static proxy under Advanced settings when creating or editing your website sync.

If your sitemap includes URLs with non-ASCII characters (such as accented letters, or scripts like Chinese or Arabic) some of those pages may not sync as expected. Sitemap discovery now supports these URLs, but they can still run into issues in other parts of the sync process. Try resyncing manually to resolve it. If pages are still missing, contact support.

When you sync website content, you may see different statuses that indicate what happened during the process. To see your website sync status go to Train > Content and select the website source, then use the Status dropdown to filter by:

Here’s what each one means and what you can do next:

The page sync is still in progress. An initial sync can take anywhere from a few minutes to over an hour based on how much content you have.

The page was successfully synced and can be enabled for Fin and Copilot.
​

These pages are intentionally not synced because you excluded them in the Advanced sync settings. They can't be retried or included unless otherwise specified.

These errors mean the sync didn’t complete and may require changes on your side before retrying:

1. Unknown error

2. Session blocked / Rate limited

3. Network, timeout, or similar errors

4. Duplicate

5. Keyword filtering

6. Status code 400

7. Blocked URL

​You can retry a failed page sync by hovering over the page, select the three dot menu and then select Resync.