Parse pdf file using chatgpt and upload the parsed file to one drive/ google drive

Sorry, this is going to be a long post. Please bear with me :nerd_face:

I want to parse multiple pdf files which contain tabular data using chatgpt and put it into an excel sheet. All pdfs are different from each other and hence i cannot specify the column names that I want to extract. All i want to do is get the table from each pdf file into an excel sheet. This my first time using make. The closest solution i came across was this and i tried to follow it. However I am not entirely sure I am doing it correctly.

I am using one drive to store my files but it isn’t mandatory. I can even use google drive.

Here’s the flow:

  1. Onedrive: Watch files by update time

  2. Onedrive: Download file (Not sure about this step because i need to select one specific file which does not help my case when there are multiple files that i want to parse)

  3. OpenAI: Upload a file

  4. OpenAI: Message an assistant

  1. OpenAI: Create a completion (Not sure if this is the correct step)

  2. Stuck here. If chatgpt can really return downloadable excel files then i wish to download it and upload it to my one drive folder. If it cannot return downloadable files then i want to retrieve the parse content and upload it to a google sheet.

I dont know how to proceed further. Please help

Welcome to the Make community!

You need to select “Knowledge Retrieval” for the assistant to be able to access files.


Additional fees apply.

1 Like

When i did that, it took a very long time to execute. I got time out error. My pdf file is not too big. It is just 1 page

Also, I didn’t get it. Additional fees apply for what?

Fees apply for using the “Retrieval” function on OpenAI.


Ohh okay.
I made the changes you suggested. I ticked knowledge retrieval.
Can you please tell me how can i transfer my parsed data that i receive from chatgpt to a google sheet? or if chatgpt can directly provide a downloadable file, i’d like to upload it directly to my one drive

1 Like