Unanswered

How can I extract text from PDF files stored in SharePoint using Microsoft Power Automate without relying on third-party OCR services, and directly provide the file to either ChatGPT or AI Builder for text extraction and summarization?

(1) Share

Report

Posted on by hammadyounas007

I'm working on a project to extract text from PDF files uploaded to a SharePoint document library using Microsoft Power Automate. However, I don't have access to third-party OCR services. Is there a way to directly upload the PDF file or its URL to either ChatGPT or AI Builder in Power Automate for text extraction and provide a summary? I know AI builder is costly so i want to use chatGPT for it. Any suggestions or guidance on achieving this using built-in actions or connectors in Power Automate would be greatly appreciated. Thank you.

Categories:

AI Builder

I have the same question (0)

All responses (6)

Answers (0)

hammadyounas007 8 on at

Like (0)

Report
Copy link

Link copied!

I'm working on a project where I need to extract text from PDF files uploaded to a SharePoint document library using Microsoft Power Automate. However, I don't have access to third-party OCR services. Is there a way to directly upload the PDF file or its URL to ChatGPT in Power Automate to extract text and provide a summary? Any suggestions or guidance on how to achieve this using built-in actions or connectors in Power Automate would be greatly appreciated. Thank you.

Was this reply helpful? Yes No
takolota1 4,980 Moderator on at

Like (0)

Report
Copy link

Link copied!

@hammadyounas007

This template OCRs a chosen PDF file, creates a replica text layer output & then passes that text to a GPT prompt.

https://powerusers.microsoft.com/t5/Power-Automate-Cookbook/Extract-Data-From-PDFs-and-Images-With-GPT/td-p/2201345

The OCR & GPT actions do consume AI Builder credits, but you should be able to atleast get a couple hundred PDFs through it with only the 5,000 AI Builder credits provided with the Power Automate premium license.

Was this reply helpful? Yes No
hammadyounas007 8 on at

Like (0)

Report
Copy link

Link copied!

@takolota I have the pdf in attachments, How can I give it AI Builder?
I have tried to make the following flow. Can you guide me How can i pass the attachment to ChatGPT?

Was this reply helpful? Yes No
takolota1 4,980 Moderator on at

Like (0)

Report
Copy link

Link copied!

@hammadyounas007

If you are trying to pass it to GPT4 Vision or GPT4o then it can only accept images. So you would need to use the Adobe connectors to convert the PDF to an array of images base64 or use a custom Azure Function to do the same. Then you could feed those images base64 to the HTTP call to GPT.

Or you could use my previous template that only pulls the PDF text with an OCR action.

Was this reply helpful? Yes No
hammadyounas007 8 on at

Like (0)

Report
Copy link

Link copied!

@takolota Can you modify your previous template to get attachment from the sharepoint not outlook? i am facing that issue on that. But I have the document which have more information not invoice

Was this reply helpful? Yes No
takolota1 4,980 Moderator on at

Like (0)

Report
Copy link

Link copied!

@hammadyounas007

There is a SharePoint Get file content action you can switch out the OneDrive action for. Then just add the dynamic content for the file content output to the OCR action.

As for setting up for a different type of document, you can change the prompt to request different data from a different document.

Also if you want to use Azure GPT4o, I just released a new template for that: https://powerusers.microsoft.com/t5/Power-Automate-Cookbook/Extract-PDF-Data-With-GPT4o/m-p/2805514#M2882

Was this reply helpful? Yes No