Using AI Builder forms with Power Pages

AI Builder in the Power Platform enables you to add intelligence to your apps without writing code. One of the most powerful use cases is integrating AI Builder Forms into Power Pages (formerly Power Apps Portals) to automate document processing, extract structured data from unstructured forms, and improve user interactions.

In this guide, we’ll walk through a step-by-step explanation of how to use AI Builder Forms with Power Pages, covering setup, configuration, customization, and deployment.

1. What is AI Builder Form Processing?

AI Builder’s Form Processing model allows you to train AI to extract text from structured documents (e.g., invoices, receipts, forms). It reads key-value pairs and tables, transforming them into usable data that can be stored in Microsoft Dataverse.

When integrated with Power Pages, users can upload forms (like PDFs or images), and the AI model extracts the data and presents it in a structured way—without needing manual data entry.

2. Prerequisites

Before you begin:

A Power Platform environment with AI Builder credits.
A Power Pages site provisioned.
Appropriate permissions to access Dataverse, Power Apps, and AI Builder.
Your structured forms (PDFs, JPEGs, PNGs) for training the AI model.

3. Step-by-Step: Create an AI Builder Form Processing Model

Step 1: Go to AI Builder

Navigate to Power Apps portal
In the left panel, select AI Builder > Explore
Under Form processing, click Create

Step 2: Train the Model

Upload at least 5 sample documents (same structure, different values)
Label fields on the documents (e.g., “Name”, “Invoice Date”, “Amount”)
Save and train the model
Once trained, publish the model

4. Create a Power Automate Flow to Trigger the Model

AI Builder works in Power Pages through automation. Here’s how:

Step 1: Go to Power Automate

In Power Automate, click + Create > Automated Flow

Step 2: Choose a Trigger

Select When a row is added (Dataverse trigger)
Connect it to a table (e.g., Form Submissions) that will store uploaded files

Step 3: Add AI Builder Action

Add AI Builder – Predict action
Select the Form Processing Model you trained
Map the file from the trigger (e.g., uploaded document)

Step 4: Handle Extracted Data

Use Update a row or Create a new row in Dataverse
Save the extracted fields into appropriate columns

Step 5: Save and Test

Save and test the flow by submitting a sample form

5. Integrate with Power Pages

Now, let’s expose this to users on your Power Pages site.

Step 1: Add File Upload Form

In Power Pages design studio:
- Add a basic form (e.g., SubmitDocument) that connects to the Dataverse table used in the flow
- Include a field to upload the file
- Optionally add user details, comments, etc.

Step 2: Customize Post-Submission Message

After submission, show a message: “Thank you. We’re processing your form.”
You can also redirect to a progress or confirmation page

Step 3: Display Extracted Data

Create a list or view in Power Pages to show users the extracted results
Link it to the same table where extracted values are stored
Optionally add a status column (e.g., “Processing”, “Completed”)

6. Enhance with Status Tracking

To improve the user experience:

Add a Status column to the Dataverse table (Processing, Success, Failed)
Update this value in the Power Automate flow
Display it as a badge or label on Power Pages for transparency

7. Security Considerations

File size & type restrictions: Limit uploads to necessary formats (e.g., PDF, JPEG)
Data loss prevention: Use Microsoft Purview DLP policies for compliance
Role-based access: Only authenticated users should upload/view sensitive data

8. Use Cases

Here are some powerful use cases of integrating AI Builder forms into Power Pages:

Student application forms: Automate form reviews for university admissions
Invoice processing: Vendors submit invoices that are parsed and reviewed instantly
Insurance claims: Customers upload claim forms and documents for quick processing
HR onboarding: Employees submit forms that are automatically processed and added to the system

9. Best Practices

Train with high-quality, consistent documents
Test your AI model with variations before publishing
Use friendly instructions on the Power Pages upload form
Provide users with feedback if extraction fails
Log data and handle failures gracefully in Power Automate

10. Maintenance and Monitoring

Retrain your model periodically if form structure changes
Monitor flows in Power Automate for errors
Set up alerts or dashboards for administrators to track status
Archive old submissions regularly to reduce storage load