Advanced data analysis lets you upload pdfs but I’m wanting a feature to let you upload pdf files with text and then basing the conversion on that upload(s). You can send text excerpts and use that going forward, but would be nice to give the GPT model more in depth background in a less cumbersome way then text
I very much agree here, this text entry only interface is extremely primitive. I’ve got annual financial statements from auditors, and I want to be able to have my assistant read the pdf ad pull out the balance sheet contained in the pdf report and then formatted as a CSV file.
I got as far as defining the method for the assistant, in the playground mode, to correctly format a CSV output of the Balance sheet, but it was painfully obvious the assistant cannot read the pdf due to a lack of OCR recognition capacity. Heck even preview.app on OSX can copy text from a .pdf
So why is thus super duper, all knowing, sentient AI unable to read a .pdf ?