Train on PDF

Supported formats

Grivo accepts standard PDF files. The text content is extracted automatically - images and charts inside PDFs are not processed.

• .pdf files (text-based PDFs work best)
• Scanned PDFs with OCR text layers are supported
• Image-only PDFs (no text layer) are not supported

Upload PDFs

1. Go to your chatbot's Training tab
2. Click Add Source → PDF Upload
3. Drag and drop your files or click to browse
4. Grivo extracts the text and adds it to your chatbot's knowledge base

You can upload multiple PDFs. Each one becomes a separate training source that you can remove or replace independently.

Best practices

• Use text-based PDFs - PDFs exported from Word, Google Docs, or Notion have clean text. Scanned documents may have OCR artifacts.
• Break large manuals into sections - A 200-page manual works, but uploading separate chapters gives the chatbot better context awareness.
• Combine with website training - Use PDFs for content not on your website (internal docs, detailed guides) and website crawl for your public site.
• Keep content current - Re-upload PDFs when the content changes. Delete outdated versions to avoid conflicting answers.

File limits

See also: Train on website · Train on docs