how to create azure openai

To use Azure OpenAI (ChatGPT-style) with Microsoft Teams, first create an Azure OpenAI resource, then build and deploy a chat experience that can be surfaced in Teams.

1. Create an Azure OpenAI resource
   1. Ensure there is an Azure subscription.
   2. Install and sign in with Azure CLI:

            az login
      

   3. Set environment variables for resource group, OpenAI service name, and region (location):

            export RESOURCE_GROUP="<group-name>"
            export OPENAI_SERVICE_NAME="<azure-openai-name>"
            export LOCATION="<azure-region>"
      

      The region must support the model to be used (for example gpt-4o or gpt-4o-mini).
   4. Create a resource group and Azure OpenAI resource (following the pattern in the tutorial):
      • Create the resource group in the chosen region.
      • Create the Azure OpenAI resource in that resource group and region.
   5. In Azure AI Foundry, deploy a chat-capable model:
      • Go to Azure AI Foundry portal.
      • Select the directory, subscription, and the Azure OpenAI resource.
      • Go to Models + endpoints → + Deploy model.
      • Choose gpt-4o or gpt-4o-mini and complete the deployment.
2. Test the chat model in the Chat playground
   1. In Azure AI Foundry, open Chat playground.
   2. Select the deployed model in the Deployment dropdown.
   3. Use the prompt area to verify that the model responds as expected.
3. Prepare to integrate with Teams There are two main patterns from the provided context: • Build a web app or service that calls Azure OpenAI
   • Use the Azure OpenAI resource from an app (for example, a web app in Azure App Service) that calls the chat completions API.
   • The .NET tutorial shows the pattern of creating an Azure OpenAI resource and then building a chatbot web app that calls it.
   • Use “Azure OpenAI on your data” and deploy directly as a Teams app (preview)
   • If using Azure OpenAI On Your Data, after configuring the chat model with data, it can be deployed as a Teams app directly from Azure AI Foundry.
4. Deploy a chat experience to Microsoft Teams (preview path)
   1. Ensure a Microsoft 365 developer or tenant account is available and that custom Teams apps and custom app uploading are enabled (per the Teams preparation guidance in the context).
   2. In Azure AI Foundry, after configuring the chat model (for example, with your data):
      • Select Deploy → a new Teams app (preview).
      • Enter the Teams app name.
      • Download the generated .zip package.
   3. Extract the .zip and open the folder in Visual Studio Code if customization is needed.
   4. Upload the app package into Teams (via the Teams client or Teams admin center, depending on tenant policy) so that users can access the chat experience inside Teams.
5. Permissions and roles
   • Ensure the account has one of the required roles on the Azure OpenAI resource, such as Cognitive Services OpenAI user, Cognitive Services OpenAI Contributor, or Cognitive Services Contributor, depending on the specific quickstart followed.

This flow gives an Azure OpenAI-backed chat experience that can be surfaced in Microsoft Teams either via a custom Teams app (preview) or via a bot/web app that calls the Azure OpenAI resource.

References:

• Quickstart: Get started using chat completions with Azure OpenAI in Azure AI Foundry Models (programming-language-studio)
• Use vision-enabled chat models (ai-foundry-portal)
• Azure OpenAI On Your Data
• Tutorial: Build a chatbot with Azure App Service and Azure OpenAI (.NET)
• Quickstart: Chat with Azure OpenAI models using your own data (programming-language-typescript)
• Quickstart: Get started using chat completions with Azure OpenAI in Azure AI Foundry Models (programming-language-typescript)
• Quickstart: Chat with Azure OpenAI models using your own data (programming-language-javascript)