Hey everyone. gpt-4o assistant, document uploads. Error on the first request. Placeholders showing in response text

API Bugs
1

Hey everyone. I’m facing a couple of issues:

  1. We’ve created a document processing platform using gpt-4o assistant. When uploading a .pdf or .docx file, the first request returns the following response: “The file does not contain visible textual content available for direct analysis. Please refine your request or ask a more specific question about the document’s content.” However, if I resend the same request, the assistant responds correctly based on the uploaded document. (Image 1). How can I make sure that the correct response comes through on the first request?
  2. When the assistant responds, it inserts placeholders in the text. I want the text to be displayed without them (Image 2). How can I make this happen?
    3
    31353×425 76.2 KB

How can I fix this?

Please help me figure out these issues.

2

Can I have one sample document? because what I believe for the text extraction it is using some python package, in most of the cases behind the scenes always some python package. Which limit the ability. Is it not possible for you to pre-process the document and then upload? This way it will also use less token.