EXTRACT INVOICE DATA WITH AI BUILDER WHEN A COPY ARRIVES IN EMAIL AND LIST THE DATA INTO DESTINATION EXCEL SHEET

[AMICHAWL2]

Hi! 

I'm a new bee in this world of Power Automate and AI with almost no technical knowledge.

I have created an AI model which works to extract the data from a PDF Invoice copy Stored in One Drive or Sharepoint and add the data in an excel file.

Now, I would like this model to extend its working and extract the same data using a flow when that invoice is arrived in the mailbox rather than stored in any folder on the system or cloud.

Can I please get this help it would be a great time-saving effort for a process to track the invoices and remove manual data entry on arrival.

In case you need any information, please feel free.

Regards,

Amit Chawla

[JoeF-MSFT]

Hello Amit - welcome to Power Automate and AI Builder!

In your flow, you will need to change the first action of the flow which is the trigger that starts the automation. Select the When a new email arrives (V3) trigger - assuming that you're using Office Outlook to receive the invoices.

In the trigger you will need to configure the following settings:

• Include Attachments: Yes
• Only with Attachments: Yes

This will trigger for every email that comes with an attachment. You might also want to configure the From or Subject Filter settings if you want the flow to run only for certain sender or when there's a specific keyword on the subject.

The flow should look like this:

Let us know if you're able to run the flow and hopefully it helps you save tons of time! 😀

[AMICHAWL2 · Accepted solution]

Thank you so much for the response. It helped me a lot to go further. Now I have a concern, Like every vendor has a different style of making an invoice, how can I get the required fields without having to train the model too much since this would be again a time taking activity looking at the basic requirement of 5 example invoices for the model to accept it successfully.

[JoeF-MSFT]

Thank you for the update @AMICHAWL2 - when training a new document processing model, I suggest you pick 'Unstructured documents' as this should give you the best results. You can also separate different vendors in different collection. The following documentation page explains how: Create a document processing custom model in AI Builder | Microsoft Learn

[takolota]

You can also now try this template using AI Builder to extract any PDF data without training different models for different formats:

https://powerusers.microsoft.com/t5/Power-Automate-Cookbook/Extract-Data-From-PDFs-and-Images-With-G...