I have uploaded a PDF into the knowledge base but it says it can’t read PDFs even though that is checked under configuration that users can upload docs. I’ve tried to reach out to supports but the contact buttons on the website aren’t working.
Couple things to check, can you highlight the text in the pdf or is the text images instead? Run your pdf through an OCR program if the text is image based, then upload to knowledge base.
Are you certain that the PDF is a modern PDF with a(n invisible to you) text layer? When PDFs originated, they were effectively images only–there wasn’t a text layer. That came to fruition around '93. If your PDF is older than that, it’s possible that the AI can’t 'read" it because it’s looking at images of text–not actual text with a text layer. Dunno if that will help, but hope it might.
Also, underneath every response in the Builder, you can click this magnifying glass and get a breakdown of exactly where your AI response is pulling from.
You can see how your knowledge base is being queried, and you can even edit the knowledge base chunks in real time (with a Pro account).