Train on your website

Chatixy starts at the URL you provide and follows internal links up to your plan’s page limit, indexing the visible text content of each page. Pages behind logins, paywalls or robots.txt blocks are skipped.

Chatixy knows your website in two layers. Every page your site lists is mapped - up to 100,000 pages per website on every plan - and the assistant opens any mapped page live when a visitor asks about it. On top of that, your key pages are learned in full: Starter 150, Pro 2,500, Business 10,000 pages per website. A daily crawl allowance caps the pages learned across *all* your websites in one day (Starter 150, Pro 12,500, Business 50,000 pages) and resets at 00:00 UTC. If a crawl is refused because the daily allowance is used up, the knowledge tab tells you so and you can run it again after the reset.

The crawl usually finishes in a few minutes for typical marketing sites; very large sites are processed in the background and your support agent improves as pages are added.

You can exclude paths you never want the support agent to learn from - careers pages, legal archives, internal tools.

• Add path patterns like /blog/archive/* in the project’s knowledge settings
• Excluded pages are removed from the index on the next crawl
• robots.txt disallow rules are always respected

You can trigger a manual re-crawl at any time from the knowledge tab - useful right after you publish new docs or change pricing. A re-crawl is available again 14 days after the last one finished.

You can also switch on scheduled re-crawls per website and pick the cadence yourself - every 14, 30 or 90 days. The same options are available on every plan.