Automating Invoice Processing Using AI Models in Power Automate

Sometimes in Power Automate, we get invoices that have a lot of information like vendor name, invoice number, date, and amount—all mixed together. But most of the time, we only need a few key details to process them or save in Excel.

In this article, we will look at a real example: whenever a new email comes in with an invoice, we will use an AI model in Power Automate to automatically pick out the important information, such as invoice id, date, vendor name, and etc. After that, we will save this information directly in an Excel file, so we don’t have to do it manually. I will show a simple step-by-step way to do this, so even if you are new to Power Automate or AI models, you can follow along easily.

End Result:

Note (Very Important):

Before we start, you must have a trained AI Invoice model. If you haven’t trained your model yet, don’t worry. I have written a detailed article on how to train an AI Invoice model. You can follow that guide first to create and train your model.

Here’s the link to the article: Train AI Models for Invoice

Prerequisites:

Basic understanding of creating flows in Power Automate.
A trained Invoicing AI model for processing documents.

Below are the steps to automate invoice processing using AI models in Power Automate:

Step 1: Create an Automated Flow

First, go to make.powerautomate.com and click on Create, then choose Automated Cloud Flow. In the trigger search, When a new email arrives (V3) and select it. Make sure to set Include Attachments to Yes, and also set Only with Attachments to Yes so that the flow only runs when an email contains attachments.

Step 2: Create an Excel File in OneDrive

Next, add a Create File action from OneDrive in your flow. This is where we will create a new Excel file to store all the invoice information that we extract from the emails.

Choose the folder path where you want to save the Excel file. For the File Name, you can give any unique name, but make sure to add the .xlsx extension. In my case, I used expression concat(formatDateTime(utcNow(),'yyyyMMddHHmmss'),'.xlsx') so that every time a new email arrives, it creates a unique file automatically. For now, keep the File Content blank by adding just a single space.

Step 3: Add a Worksheet to the Excel File

After creating the Excel file, add the Create Worksheet action. For the Location, select OneDrive for Business, and for the Document Library, select OneDrive. In the File field, choose Id from the dynamic content of the Excel file which we created in Step 2. Finally, give any name you like for the worksheet; in my example, I used InvoiceDetails.

Step 4: Create a Table in the Worksheet

Add the Create Table action. For Location, select OneDrive for Business, and for Document Library, select OneDrive. In the File field, choose Id from the dynamic content of the Excel file we created in Step 2.

For the Table range, it’s important to use the format 'WorksheetName'!TableRange. In this example, the worksheet name is InvoiceDetails and I want to show 7 columns, so the range is 'InvoiceDetails'!A1:G1. Adjust this range if your Excel table has more columns.

In the Table name field, give any name you like; in my example, I used Table1. Finally, under Columns names, list all the columns you want to include in the table after extracting the information.

Step 5: Get Email Attachments

Add an Apply to each action and set the output to Attachments from the trigger.

Inside the Apply to each, add the Get Attachment (V2) action. Here, you need to map two fields:

Message Id : select Message Id from the dynamic content.
Attachment Id : select Attachments Attachment Id from the dynamic content.

Step 6: Process the Attachments with AI Model

Inside the same Apply to Each loop after the Get Attachments action, add the Process Documents action. For the AI Model, select the model you have trained for your documents. Set the Form Type to PDF Document, and in the Form field, choose the dynamic content Content Bytes from the Get Attachments action.

Step 7: Add Data to the Excel Table

Inside the same Apply to Each loop, add the Excel action Add a Row into a Table. For Location, select OneDrive for Business, and for Document Library, select OneDrive. In the File field, pass the Id from the Create File action in Step 2. For the Table, select the dynamic content name from the Create Table action.

At the end, define the row fields object manually. Make sure to include only the columns you added in Step 4, and copy the exact column names. Then map the fields from the Process Documents action to match your requirements.

The row fields must be written in a valid JSON format, enclosed within {}.

Conclusion

By following these steps, you can automatically extract data from email attachments using AI and store it in Excel. This helps save time, reduce manual work, and ensure your data is processed accurately and efficiently.