Configure document processing by using AI Builder

Step 1: Create a New Flow

Create a new flow in Power Automate and select the trigger that initiates the document processing. For example, you can use the “When a file is created or modified (properties only)” trigger to detect changes in a document library.

Step 2: Add AI Builder Action

Add an action to the flow and search for “AI Builder” in the actions list. Select the “Analyze a document (preview)” action.

Step 3: Establish AI Builder Connection

In the AI Builder action settings, click on the “Add new connection” button to establish a connection with AI Builder. This step is necessary to access AI models and perform document analysis.

Step 4: Select Environment

Once the connection is established, select the environment associated with your AI Builder service. If you haven’t created an environment yet, refer to the AI Builder documentation for instructions on setting up an environment.

Step 5: Configure Document Analysis

Next, configure the AI Builder action by specifying the document you want to analyze. You can choose to analyze a file from SharePoint, OneDrive, or any other cloud storage location supported by Power Automate.

Step 6: Provide Document Details

Provide the necessary details for document processing, such as the file path or URL. You can also specify whether you want the action to output just the document text or extract key-value pairs from the document.

Step 7: Select AI Builder Model

If you want to extract key-value pairs, you need to use an AI Builder model that is trained for entity extraction. In the “Model analysis kind” dropdown, select the relevant model for your document processing needs.

Step 8: Specify Page Range

Optionally, you can limit the output to a specific page range by specifying the page numbers in the “Page range” field. This is useful when dealing with multi-page documents.

Step 9: Add Subsequent Steps

After configuring the AI Builder action, you can add subsequent steps in the flow to perform additional tasks based on the document analysis. For example, you can use conditional statements to check for specific values extracted from the document and trigger different actions accordingly.

Step 10: Test and Save

Save the flow and test it by uploading a document that matches your trigger conditions. Power Automate will automatically analyze the document using the AI Builder model and perform the specified actions based on the analysis results.

Here’s an example of how you can configure document processing using AI Builder in Power Automate:

{
"triggers": {
"When_a_file_is_created_or_modified_properties_only": {
"inputs": {
"host": {
"connectionName": "shared_sharepointonline_12345678",
"operationId": "When_a_file_is_created_or_modified_properties_only"
},
"parameters": {
"folderId": "",
"skipToken": ""
}
},
"recurrence": {
"frequency": "Minute",
"interval": 15
}
}
},
"actions": {
"Analyze_a_document_(preview)": {
"inputs": {
"host": {
"connectionName": "",
"operationId": "Analyze_a_document_(preview)"
},
"parameters": {
"fileUrl": "",
"modelAnalysisKind": ""
}
}
},
"Condition": {
"actions": {
"If_true": {
"actions": [
{
"method": "GET",
"uri": ""
}
],
"expression": "@equals(outputs('Analyze_a_document_(preview)').body.'" +
"" +
"', '')"
}
},
"expression": "@contains(toLower(outputs('Analyze_a_document_(preview)').body.text), '')"
}
}
}


Note: The code above is just an example, and you need to replace the placeholder values with the actual values corresponding to your AI Builder connection, document location, model name, and desired actions.

By following these steps and customizing the flow to your specific requirements, you can configure document processing using AI Builder in Microsoft Power Automate.