Vector Search

Vector Search finds topics by meaning, not only by matching words. A reader who searches for "how do I rotate a credential" finds your topic "Changing your API key", even though the two share no words. It powers both the search box in your published documentation and AI answers.

Category: Organization · Plan: Business or higher · Requires: nothing

Index your existing content

After you install the addon, topics you publish are indexed automatically. Content published before that must be indexed once:

1. Open Organization Settings > Search. Until existing content is indexed, the tab shows "Existing content is not indexed yet".
2. Select Index all content.
3. Type INDEX to confirm, then select Start indexing.
4. Keep the browser tab open while indexing runs. The tab shows how many topics are done so far.

If indexing stops, for example because you closed the tab, select the button again and it continues where it stopped. When it finishes, the tab shows how many entries are indexed and when. If some topics couldn't be indexed, run it again to retry them.

Indexing a large knowledge base can take a while. To rebuild the whole index later, select Re-index all content.

If you uninstall it

Uninstalling deletes your organization's search index. Search results for content published before the uninstall can be missing until the content is indexed again. If you reinstall the addon, run Index all content again.