How to Use AI Builder in Power Platform

AI Builder is a powerful tool within the Microsoft Power Platform that allows you to add artificial intelligence (AI) capabilities to your apps and workflows without needing extensive coding or data science expertise. Here’s a step-by-step guide on how to use AI Builder in Power Platform:

1. Understand AI Builder Capabilities

AI Builder provides prebuilt and custom AI models for various use cases, including:

Form Processing: Extract data from forms and documents.
Object Detection: Identify and count objects in images.
Prediction: Predict outcomes based on historical data.
Text Classification: Categorize text data.
Sentiment Analysis: Analyze text sentiment (positive, negative, neutral).
Language Detection: Detect the language of text.
Business Card Reader: Extract contact information from business cards.

2. Prerequisites

Power Platform Environment: Ensure you have access to a Power Platform environment with AI Builder enabled.
Licensing: AI Builder requires a license (per-user or per-flow plan).
Data: Prepare the data needed to train your AI model (if using custom models).

3. Steps to Use AI Builder

Step 1: Choose an AI Model

Go to the Power Apps or Power Automate portal.
Navigate to AI Builder in the left-hand menu.
Select Build a Model.
Choose the type of AI model you want to create:

Prebuilt Models: Ready-to-use models for common scenarios.
Custom Models: Train your own model using your data.

Step 2: Create and Train a Custom Model

Select a Model Type:

For example, choose Form Processing to extract data from invoices or receipts.

Upload Training Data:

Provide sample documents or data to train the model.
For form processing, upload at least 5-10 sample forms.

Tag Data:

Manually tag the fields you want the model to extract (e.g., invoice number, date, total amount).

Train the Model:

Click Train to train the model using your tagged data.
Training time varies based on the model type and data size.

Test the Model:

Use the Quick Test feature to test the model with new data.
Review the results and refine the model if necessary.

Step 3: Publish the Model

Once the model is trained and tested, click Publish.
The model will be available for use in Power Apps and Power Automate.

Step 4: Use the Model in Power Apps

Open your Power App in the Power Apps Studio.
Add the AI Builder model to your app:

Go to the AI Builder section in the left-hand menu.
Select your published model.

Configure the model to process data:

For example, use a form processing model to extract data from uploaded documents.

Bind the model’s output to controls in your app (e.g., display extracted data in labels or text boxes).

Step 5: Use the Model in Power Automate

Create a new flow in Power Automate.
Add the AI Builder action:

Search for your published model in the actions list.

Configure the action:

For example, use a form processing model to extract data from an email attachment.

Use the extracted data in subsequent steps (e.g., save to SharePoint, update a database).

4. Examples of AI Builder Use Cases

Form Processing

Extract data from invoices, receipts, or surveys.
Automate data entry into systems like SharePoint or Dynamics 365.

Object Detection

Count items in inventory using images.
Detect defects in manufacturing processes.

Prediction

Predict customer churn based on historical data.
Forecast sales or demand.

Sentiment Analysis

Analyze customer feedback or social media comments.
Route negative feedback to support teams.

Business Card Reader

Automatically extract contact details from business cards and save them to a CRM.

5. Best Practices for Using AI Builder

Start Small: Begin with prebuilt models to understand AI Builder’s capabilities.
Use Quality Data: Ensure your training data is accurate and representative.
Iterate and Improve: Continuously test and refine your models for better accuracy.
Monitor Usage: Track model performance and usage to optimize costs.
Combine with Other Power Platform Tools: Use AI Builder with Power Apps, Power Automate, and Power BI for end-to-end solutions.

6. Licensing and Costs

AI Builder Credits: Each model consumes credits based on usage.
Plan Types:
Per-User Plan: For individual users.
Per-Flow Plan: For automated workflows.
Free Tier: Limited usage is available for testing and exploration.

7. Resources and Learning

Microsoft Documentation: Refer to the official AI Builder documentation.
Community Forums: Join the Power Platform community for tips and support.
Training: Explore Microsoft Learn for AI Builder tutorials and certifications.

By following these steps and best practices, you can leverage AI Builder to add intelligent capabilities to your Power Platform solutions, automate processes, and gain insights from your data.