In today’s data-driven business environment, organizations constantly handle a wide range of documents—such as invoices, receipts, registration forms, loan applications, and more. Traditionally, extracting data from these forms is a manual, error-prone task that consumes valuable time and resources. Enter AI Builder Form Processing, a powerful, low-code feature within the Microsoft Power Platform that automates the extraction of structured data from forms using artificial intelligence (AI).
This article explores how AI Builder Form Processing works, its key features, real-world use cases, implementation steps, integration scenarios, and best practices to get the most out of this intelligent automation tool.
What is AI Builder?
AI Builder is a no-code/low-code artificial intelligence platform embedded in Microsoft’s Power Platform—including Power Apps, Power Automate, and Power Pages. It allows business users and developers to infuse AI capabilities into apps and workflows without writing complex machine learning code.
AI Builder offers models for:
- Form processing
- Object detection
- Prediction
- Sentiment analysis
- Invoice processing
- Receipt scanning
- Text recognition (OCR)
One of its most practical and widely used features is Form Processing, which leverages AI to extract data from structured or semi-structured documents.
What is Form Processing in AI Builder?
Form Processing in AI Builder uses Optical Character Recognition (OCR) and machine learning models to extract key-value pairs and table data from documents such as PDFs, images, and scanned forms.
Instead of relying on predefined templates or hard-coded rules, the AI model learns from sample documents and adapts to new layouts within a similar structure. It then identifies and extracts the relevant fields—such as names, dates, amounts, or addresses—so that this information can be used downstream in business processes.
Key Features
Here are the standout features of AI Builder Form Processing:
1. Low-Code Model Training
Users can upload sample documents and label fields visually using a drag-and-drop interface—no coding required.
2. Support for PDFs and Images
Supports common file formats like PDF, PNG, JPG, and TIFF.
3. Table Extraction
Detects and extracts repeating rows of data within forms, such as line items in invoices or product lists in order forms.
4. Flexible Layout Detection
Handles variations in layout, as long as the data structure is generally consistent.
5. Integration with Power Platform
Models can be used in Power Automate flows, Power Apps, Power Pages, or embedded in Dynamics 365 applications.
6. Secure and Scalable
Form data is processed in compliance with Microsoft security and compliance standards, including GDPR and HIPAA.
Common Use Cases
1. Invoice Processing
Extract vendor name, invoice number, due date, line items, and total amount from invoices and feed them into ERP or accounting systems.
2. HR Onboarding Forms
Capture employee data from scanned forms, such as job applications, ID verification, and benefit enrollment documents.
3. Loan Application Review
Pull structured data from submitted PDFs to automate decision-making processes in financial services.
4. Insurance Claims
Extract claimant details, policy numbers, and descriptions from claim forms for faster claims processing.
5. Healthcare Intake Forms
Digitize patient intake and consent forms for EHR (Electronic Health Record) integration.
How AI Builder Form Processing Works
Let’s walk through the process of creating and deploying a form processing model in AI Builder.
Step 1: Prepare Your Documents
- Gather at least five sample forms with similar structure (e.g., purchase orders).
- Ensure clarity in scanned documents for optimal OCR performance.
Step 2: Create a Form Processing Model
- Go to Power Apps or Power Automate portal.
- Select AI Builder > Explore > Form Processing.
- Name your model and upload your sample documents.
Step 3: Tag Fields
- Use the visual editor to draw boxes around fields you want the AI to extract (e.g., “Invoice Date”, “Total”).
- Repeat the tagging across multiple forms to improve accuracy.
- Save and train the model.
Step 4: Test the Model
- Upload a new document and verify the extracted data.
- Make adjustments if needed and retrain.
Step 5: Use the Model in an App or Flow
- Integrate the model in a Power Automate flow to automatically trigger on new documents.
- Embed it into a Canvas App to allow users to upload forms and see extracted data live.
- Feed the results into Dataverse, SharePoint, or third-party APIs.
Example Scenario: Automating Invoice Data Entry
Let’s look at how a finance department can use AI Builder to eliminate manual invoice data entry.
Problem:
Staff manually extracts vendor and invoice data from incoming PDF invoices and enters them into Dynamics 365 Finance.
Solution:
- Train a Form Processing Model using 10–15 past invoices.
- Build a Power Automate Flow:
- Triggered when a new invoice is added to a SharePoint folder.
- Calls the trained AI Builder model to extract vendor, invoice date, due date, and amount.
- Creates a new invoice record in Dynamics 365 Finance.
- Log and Monitor exceptions for human review.
Benefits:
- Saves time and reduces human error.
- Accelerates accounts payable cycle.
- Frees staff to focus on higher-value tasks.
Best Practices for Accuracy and Performance
1. Use High-Quality Documents
Ensure scanned images or PDFs are clean, legible, and consistent in structure.
2. Provide Enough Samples
Use at least 5–10 varied samples per form type to help the model generalize across minor variations.
3. Use Field Consistency
Tag the same fields across all documents during training—even if a field is blank or not visible.
4. Use Tables for Line Items
Use table tagging when processing invoices or forms with line items to avoid misaligned data.
5. Retrain with Errors
Periodically retrain the model with incorrectly processed forms to improve accuracy over time.
Integration Options
1. Power Automate
Use flows to automate document handling:
- Trigger from SharePoint, OneDrive, email, or Dataverse.
- Extract data using AI Builder.
- Send data to databases, approval flows, or alerts.
2. Power Apps
Embed form processing in Canvas Apps:
- Allow users to take a picture or upload a document.
- Instantly extract and display relevant information.
3. Dynamics 365
Trigger AI Builder models from within Dynamics 365 CE (Customer Engagement) workflows:
- Auto-populate case forms, quotes, or customer records.
- Streamline ticketing and document-based workflows.
Limitations to Consider
- Form layout must be consistent – models struggle with completely unstructured documents.
- Tables must be clearly defined – poorly formatted tables can reduce accuracy.
- Limited to 500 pages per document and certain file types.
- Not suitable for handwritten forms unless processed with a separate handwriting recognition model.
Licensing and Pricing
AI Builder is available as an add-on to Microsoft Power Platform and Dynamics 365 environments. It’s licensed based on AI Builder credits, which are consumed when models are run.
Microsoft offers:
- A free trial with limited credits.
- Additional credits via Power Platform per-user or per-app plans.