Enhancing Power Apps with AI: Prebuilt and Custom AI Models

AI Models

AI Builder provides a palette of models that are ready to use, and in some cases, adapted to your organization. These models can greatly improve the Power Apps user experience with interactive controls.

Prerequisites

• A Microsoft Power Apps license
• Access to Microsoft Dataverse
• An AI Builder add-on or trial

AI Builder Prebuilt Models

Prebuilt models are based on AI data and patterns that Microsoft has introduced in their training. These ready-to-use models cover the broadly available data and concepts that aren't specific to your organization or scenario.

Examples of available prebuilt models include:

• Invoice processing
• Text recognition
• Sentiment analysis
• Receipt processing
• Entity extraction
• ID reader
• Key phrase extraction
• Business card reader
• Category classification
• Text generation
• Language detection
• Text translation
• Image description

AI Builder Custom Models

Custom models require training and validation before you can use them in Power Apps or Microsoft Power Automate. They are designed for organization-specific scenarios.

Examples of custom models include:

• Prediction
• Object detection
• Category classification
• Form processing

Example 1: Sentiment Analysis Model

The Sentiment Analysis model allows us to analyze the tone of user input and classify it as positive, negative, or neutral.

Implementation Steps:

• ✔ TextInput – User enters text here.
• ✔ Button – Triggers sentiment analysis.
• ✔ Label – Displays the sentiment result.

Button OnSelect Property:
Set(sentimentResult, AIBuilder.SentimentAnalysisV3(TextInput1.Text))

Label Default Property:
sentimentResult.Sentiment

Example 2: Language Detection Model

The Language Detection model identifies the language of a given text input.

Implementation Steps:

• ✔ TextInput – User enters text.
• ✔ Button – Detects language.
• ✔ Label – Displays detected language.

Button OnSelect Property:
Set(languageResult, AIBuilder.LanguageDetection(TextInput1.Text))

Label Default Property:
languageResult.Language