Plan requirement
| Subscription | Suite Team or higher, Knowledge Professional or higher |
| Access | Admin |
Yes, through connected sources and the web crawler. But it imports rather than searches: the agent answers from a copy.
What can it reach?
- A crawled website, such as your public documentation or product pages.
- Connected systems, where a knowledge tool you already use is linked in.
Both make that content available to the AI agent as a source of answers.
It imports, it does not search
The distinction that matters. The AI agent does not search live data in a website, file or help centre: content is imported, once or recurringly, and the agent answers from that copy.
So it cannot search the open internet, which is what makes it usable. It also means a source can be out of date without looking it: the crawler refreshes every 30 minutes, and a crawl is capped at five hours by default.
Should I connect anything?
Would you be comfortable with the AI agent quoting that content to a customer? That is what connecting it means. Marketing pages promise things support cannot deliver, and old blog posts describe the product as it was. Both look authoritative in an answer.
Scope narrowly
Point the crawler at the documentation path, not the whole site. That decision determines whether this helps or produces confident answers from a page written for another purpose.
There is room to be precise: up to 50 crawlers, and 500 URLs when limiting one to individual pages. Several narrow crawlers beat one broad one.
Your articles compete with it
Every external source is another candidate answer. If your articles start losing to external pages, narrow the source or improve the article.
Comments
0 comments
Article is closed for comments.