Can you read office files with AI Builder?

(0) Share

Report

Posted on by Germán Microsoft Employee

Hello everyone!

I know that, for intelligent document processing (invocices, forms, etc), we can use AI Builder in Power Automate to extract the data from PDFs. I have been trying to search for some documentation or videos on this matter, but I haven't really found much helpful. What I was wondering:

Can the same be done with native Office files like Word (.docx)? Or do we need to convert Word files to PDF first?

I am trying to read attachments from emails, and I would like to do as least steps as possible. I know that in Azure AI this could be done, but I am not very experience with this, so I would like to avoid this.

Any idea on what would be the best (and simple) approach here?

Thanks a lot in advance! :)

Categories:

AI Builder

I have the same question (0)

All responses (2)

Answers (0)

Sort by

Suggested answer

Riyaz_riz11 4,150 Super User 2026 Season 1 on at

Like (0)

Report
Copy link

Link copied!

Hi,

AI Builder’s document processing models only support PDF, JPG, and PNG as input; Word .docx isn’t accepted directly.

So for AI Builder form or document processing you must first convert Word files to PDF e.g. using OneDrive Convert file action.

If you only need plain text not structured fields, you can avoid AI Builder and extract text from .docx by reading the internal document.xml with Power Automate and XPath.

Azure Document Intelligence does support .docx, but that’s a different service and setup from AI Builder

If I have answered your question, please mark it as the preferred solution ✅ . If you like my response, please give it a Thumbs Up 👍.
Regards,
Riyaz

Was this reply helpful? Yes No
Germán Microsoft Employee on at

Like (0)

Report
Copy link

Link copied!

Hello, Riyaz:

I really appreciate your response. I was aware that AI Builder only supports those formats, hence why my question asking if there's a way to extract the data from word documents. I thought one alternative would be what you mentioned: converting Word to PDF, but then I would have to reconvert it back in case the file must be resent

About what you mentioned, that could be a good idea! Extracting just the plain text with PA and XPath, but would that work for invoices, contracts, etc?

Thank you

Was this reply helpful? Yes No