Skip to content

Knowledge

It reads your site, then keeps reading it

The quality of every answer comes down to one thing: what the bot actually knows. ChatFlint builds that knowledge from the content you already have, and maintains it without you thinking about it.

Who this is forSites with more content than anyone can search — a long manual, hundreds of products, or five years of posts.

  • Six sources — WordPress pages, posts and products; a crawl of any URL; an XML sitemap; PDF and Word uploads; hand-written Q&A pairs; and free text you paste in.
  • Page-builder aware — Elementor, Divi and Gutenberg layouts are read properly, with headings, lists and tables intact, instead of arriving as one flat blob.
  • Section-labelled chunks — a long page is stored as “About us — Pricing”, so an answer can cite the part it came from.
  • Free to keep in sync — unchanged content is never re-processed, so a nightly sync on a 500-page site costs nothing.
  • A nightly freshness check — every synced page is fingerprinted and re-indexed the moment it changes, which catches the edits WordPress never announces: an import script, a plugin writing a field, a sale that starts by itself at midnight.
  • Cleans up after itself — a duplicate scanner, a stale-fact watch when a page loses a detail it used to state, and a knowledge reset scoped by source.

Add it in one guided flow

The Add Knowledge screen asks where your content lives and does the rest: pick post types (all of them, or the ten pages that matter), paste a URL to crawl, drop in PDFs, or type Q&A pairs. Crawls skip tag, category and shop-filter noise by default, and any URL can be excluded permanently.

See exactly what the bot sees

Every page has a “preview as the bot sees it” view, so you can check the extraction before you trust it. A retrieval tester shows which passages a question actually pulls, which is usually the fastest way to fix a bad answer.

Questions

Questions

Do I have to re-upload content when I edit a page?

No. Published changes are picked up automatically, and only the pages that really changed are re-processed.

What about content behind a login or in a custom field?

Custom fields and ACF content are included in the sync. Anything the crawler cannot reach without logging in should be added as a file or a Q&A pair instead.

Put it on your site this week

One plugin, your own content, and seven days on our AI before you add a key. No subscription to a chat platform, no data leaving your hosting.

  1. Download and install. Upload it under Plugins → Add New, like any other plugin.
  2. Start the trial. The setup wizard asks for an email; the answers run on our AI while you decide.
  3. Sync your content. One click reads your pages, posts and products. From then on it answers your visitors.

Currently at version 1.59.0 · what changed

Talk to the person who built it

Pre-sales questions, licences, agency work, or a site that needs a look. One inbox, read by the developer.

  • A reply within a working day — usually much sooner.
  • Support by the person who wrote the code, not a queue.
  • A 14-day refund on every plan, no questions that matter.