Add knowledge to the knowledge base

Four ways to teach your chatbot: manual entry, website import, file upload and bulk import from a file.

Last verified: 2026-10-20

In the “Knowledge” area you add content in four ways: as individual entries by hand, by automatically importing your website, by uploading documents, or by bulk import from a CSV, Excel or JSON file. Your chatbot uses approved content for its answers as soon as “Enable RAG” is switched on in the chatbot configuration.

Step by step

  1. Option 1: create an entry by handIn the “Knowledge base” tab, click “New entry”, enter title and content, optionally category and tags, and save. New entries are first saved as drafts.
  2. Option 2: import your websiteIn the “Import website” tab, enter your website’s address and click “Add website”. Zentor App reads your pages, respects your robots.txt and updates the content regularly and automatically. Click “Crawl now” to start a run immediately; for each page you decide whether it is active in the chatbot.
  3. Option 3: upload filesIn the “Files” tab you upload documents of up to 100 MB: PDF, Word (DOC, DOCX), Excel (XLS, XLSX), text and HTML files as well as WordPress exports. The text is taken over automatically – for Excel per worksheet, for WordPress exports per post. The status on the file shows whether it has been imported.
  4. Option 4: bulk import from a fileIn the “Knowledge base” tab, click “Import” and choose a CSV, Excel (XLSX) or JSON file – one row is one entry. The title and content columns are required; category, tags (separated by semicolon or comma) and status (approved = immediately available to the chatbot, default; draft = draft; archived = archived) are optional. A preview shows every row and any errors; entries with an existing title are skipped. Up to 5 MB and 2,000 rows are possible per file. A CSV template is available in the import dialog.
  5. Check the resultThe “Source status” tab shows how many approved sources your chatbot uses. In the “Chatbot test” tab you ask a question and see the answer and the sources used.
  6. Switch on the knowledge base in the chatbotIn the “Chatbot” area, in the “Configuration” tab, check that “Enable RAG” is switched on.

Expected result

Your content is available to the chatbot, and you have checked in the chatbot test that it uses the right sources for its answers.

Common error cases

An entry created by hand does not appear in the chatbot’s answers.

Cause: Entries newly created by hand start as drafts; the chatbot only uses approved entries.

Resolution: Create entries that should be used immediately via import – the status approved is preset there. To approve existing drafts, contact Zentor.

The import preview marks rows as faulty or skipped.

Cause: Required columns are missing, a value is too long, the title already exists or your quota of knowledge entries has been reached.

Resolution: Correct the marked rows in your file and upload it again. You can get a higher quota through Zentor.

An uploaded file is not imported.

Cause: The file is password-protected, damaged or contains no readable text, for example a scanned document or an image.

Resolution: The status on the file states the reason. Save the document again without a password or with readable text, or create the content as an entry.

Not all pages of my website were imported.

Cause: Pages blocked by your robots.txt are skipped. Very large websites are read in several runs.

Resolution: Check your robots.txt for unintended blocks. Start another run with “Crawl now” or wait for the next automatic run.

The chatbot does not use the knowledge base at all.

Cause: “Enable RAG” is switched off in the chatbot configuration.

Resolution: Switch on “Enable RAG” in the “Chatbot” area in the “Configuration” tab (see the article “Configure your chatbot”).

Related articles

← Back to category