Chatwoot
en
Language
en English pt_BR Portuguese (Brazil) es Spanish de German fr French ar Arabic

Add documents to Captain's knowledge

Open in
View as Markdown Open in ChatGPT Open in Claude
harishmita

harishmita

Last updated on Oct 20, 2026

Discover

Documents teach your assistant about your product. Add website pages or PDF files, and Captain reads them, answers from them, and creates FAQs from them.

Plan & access: Startups, Business or Enterprise Pricing & Plans. Administrators add, refresh and delete documents. Every paid plan includes up to 2,000 documents. The limit is shared by all assistants in the account, and each PDF or crawled web page counts as one document.

Set up & try

Before you start

  • An assistant. See Set up a Captain assistant and connect your inboxes.

  • Website pages must be public. Captain cannot read pages behind a login.

  • PDFs must be 10MB or smaller and contain real text (scanned images may not be readable).

  • If your website blocks crawlers in robots.txt, allow Chatwoot first (below).

Add a website or PDF

  1. Go to Captain > Documents.

  2. Click Create a new document.

  3. Choose the Document Type:

    • URL: enter the page address. Captain reads that page and the pages linked under the same path. For example, https://example.com/help covers every page that starts with that address.

      • PDF File: click Choose PDF file and pick the file.

  4. Optional: add a Document Name (Optional).

  5. Choose the assistant and click Create.

Captain reads and indexes the content. Pages it finds while crawling are added as separate documents.

Allow Chatwoot to crawl your site

Chatwoot uses Firecrawl to read websites, and it follows your robots.txt. If some pages are blocked, add this to https://yourdomain.com/robots.txt:

User-agent: FirecrawlAgent
Allow: /

Check the file is live at that address, then add or refresh the document.

If your robots.txt blocks the crawler when you add a website, no pages are added and no error is shown. The assistant then gives incomplete or generic answers.

Test it

  1. Open the document's menu and choose View details.

  2. Check Crawled content (or PDF details) shows your text, and Related FAQs lists sensible questions.

  3. Ask a question from that page in the assistant's Playground.

Get value

  • Keep documents current: use Refresh now on one document, or select several and click Refresh. The list shows when each was last updated.

  • Website documents also refresh automatically: monthly on Startups (on the 1st), weekly on Business (on Sundays) and daily on Enterprise. Processing can take longer, so a page may not refresh at that exact time. Uploaded PDFs are not refreshed automatically.

  • One website crawl adds up to 500 pages, and never more than your remaining document allowance.

  • Prefer focused sources, such as your help center, over a whole marketing site.

  • Used in conversations shows which conversations used a document.

  • Deleting a document removes everything Captain learned from it.

  • Help Center articles are not read automatically. To use them, add your Help Center's address as a URL document. Each language version of an article is a separate page, so each counts as a document.

  • FAQs created from a document go live straight away. They are listed under FAQs and need no approval.

  • FAQs that Captain generates are always in your account's language. The assistant still replies in the customer's language.

  • Captain does not remove personal or confidential details from what you add. Only add content you are happy for the assistant to quote to customers.

Troubleshoot

Symptom Likely cause What to do
Status shows Failed to sync with Access denied The page needs a login, or your site blocked the request (error 401 or 403). Make the page public, or allow the request on your site.
Added a website but no pages appear, or answers are generic Your robots.txt may block the crawler. Allow it as shown in Allow Chatwoot to crawl your site, then add the document again.
Page not found URL is wrong or the page moved Fix the URL and add it again
Page took too long to respond Your site was slow Use Retry refresh later
Page returned empty content Content loads only with scripts, or the page is empty Use a page with plain text, or upload a PDF
update stalled A refresh did not finish Use Retry refresh
PDF details says readable content could not be extracted Scanned or image-only PDF Upload a text PDF
Banner: Document limit reached The account has 2,000 documents across all its assistants Delete documents you no longer need, or see Pricing & Plans

More causes: Troubleshoot Captain.