Add knowledge to the knowledge base
Three ways to teach your chatbot: manual entries, a website crawl, or file upload.
Last verified: 2026-07-20
In the "Knowledge" section you can add content in three ways: manual knowledge entries, automatically importing your website, or uploading files. The chatbot uses this content for answers as long as "RAG Enabled" is turned on in the chatbot configuration.
Step by step
- Option A: Create a manual entryIn the "Knowledge base" tab, create a new entry with a title and content (free text) and save it.
- Option B: Start a website crawlIn the "Import website" tab, enter your website URL and start the crawl. A crawl reads up to 300 pages (max. 8 minutes total duration) respecting your robots.txt, and then repeats automatically every 24 hours. Progress updates every 4 seconds.
- Option C: Upload a fileIn the "Files" tab, upload a file (PDF, images, text/CSV, WordPress export — max. 100 MB). Searchable text content is extracted automatically.
- Make sure RAG is usedIn the "Chatbot" section → "Config", check that "RAG Enabled" is turned on so the chatbot actually uses this content for answers.
Expected result
Your content is available to the chatbot for answering customer questions once RAG is enabled.
Common error cases
A manually created knowledge entry does not seem to be used by the chatbot.
Cause: Newly created manual entries initially receive an internal draft status; there is currently no button in the tenant dashboard to change this status yourself.
Resolution: Contact Zentor if a manual entry permanently does not appear in chatbot answers.
An uploaded Word or Excel file is not searched.
Cause: The upload accepts DOC/DOCX/XLS/XLSX as file types, but automatic text extraction currently only supports PDF, HTML, plain text and WordPress exports (WXR).
Resolution: Convert the content to PDF or text and upload it again, or create the content as a manual knowledge entry.
The website crawl stops before the last page or does not capture all subpages.
Cause: A crawl is limited to a maximum of 300 pages and 8 minutes total duration; pages blocked by your robots.txt are also respected.
Resolution: Check your robots.txt for unwanted blocks; very large websites may not be fully captured in a single run — the next automatic crawl runs again after 24 hours.
The chatbot does not use the knowledge base at all.
Cause: "RAG Enabled" is not turned on in the chatbot configuration — this switch is off by default.
Resolution: Enable "RAG Enabled" in the "Chatbot" section → "Config" (see "Create and configure a chatbot profile").