Using Azure OpenAI in a Logic App Workflow

Azure Logic Apps provide a low-code way to automate workflows, integrating with various services like Azure OpenAI to enable AI-driven automation. In this article, we'll create a Logic App that leverages Azure OpenAI for text generation using GPT models.

Prerequisites

Before you begin, ensure you have,

• An Azure subscription
• An Azure OpenAI resource (GPT-4 or GPT-3.5-turbo deployed)
• An API Key and Endpoint from the Azure OpenAI resource
• Access to the Azure Portal

Step 1. Set Up Azure OpenAI.

• Go to Azure Portal and search for Azure OpenAI.
  
• Click Create and deploy the required GPT model.
  
• Navigate to the Keys and Endpoint section and copy the API Key and Endpoint.

Step 2. Create a Logic App.

• In Azure Portal, search for Logic Apps.
• Click Create and choose Consumption Plan.
• Select a Resource Group, enter a Logic App name, and choose a Region.
• Click Review + Create, then Create.
  

Step 3. Design the Logic App Workflow.

Add an HTTP Trigger

• Open the Logic App Designer.
• Choose "When an HTTP request is received" as the trigger.
• Click "Use sample payload" and enter the following JSON.

  {
      "question": "What is AI?"
  }
  

Call Azure OpenAI API

• Click "New Step" and add an HTTP Action.
• Configure it as follows.
  • Method: POST
  • URL: {your Azure OpenAI endpoint}/openai/deployments/{deployment_name}/chat/completions?api-version=2024-02-01
  • Headers
    • Content-Type: application/json
    • api-key: {your_api_key}
  • Body

{
  "messages": [
    {
      "role": "system",
      "content": "You are an AI assistant."
    },
    {
      "role": "user",
      "content": "@{triggerBody()['question']}"
    }
  ],
  "max_tokens": 100
}

Return AI Response

• Click "New Step" and add "Response".
• Set Status Code to 200.
• In the Body field, add @body('HTTP')?['choices'][0]['message']['content'].
  

Step 4. Test the Logic App.

• Click "Save," click Run, and select Run with payload.
• Add {"question": "What is AI?"} in the body.
  
• The response will contain AI-generated text.
  

Use Cases

• Automated AI Chatbot for Teams, Slack, or web apps.
• Email Automation using AI-generated responses.
• Content Summarization for customer support and research.
• Sentiment Analysis for analyzing customer feedback.

Conclusion

By integrating Azure OpenAI with Logic Apps, we can build powerful AI-driven workflows without writing complex code. This approach allows automation of various business processes, enhancing productivity and efficiency.