Prepare the Environment
Teams Copilot Starter app requires Azure OpenAI service. Before deploying to Azure, configure Azure OpenAI related variables in your .env
file. Create your Azure OpenAI resource on azure portal and deploy completion and embedding models on that Azure OpenAI resource.
OPENAI_KEY=<OpenAI Key>
OPENAI_ENDPOINT=<OpenAI Endpoint>
OPENAI_MODEL=<completetion model>
OPENAI_EMBEDDING_MODEL=<embedding model>
Open Teams Toolkit in Visual Studio Code, and sign in your Azure account by clicking the Sign in to Azure
in the ACCOUNTS
section from sidebar.
After you signed in, select a subscription under your account. The Teams Toolkit will use this subscription to provision Azure resources to host you app.
Run the teamsapp provision
command to trigger the provision stage.
teamsapp provision --env dev
Note: Provision Azure cloud resources and deploy to Azure may cause charges to your Azure Subscription.
After the Teams Copilot Starter has been deployed to Azure, the development/test and production environment architecture may look like in the following diagram:
dev.botframework.com
and token.botframework.com
for authentication tokens. Azure blobStorage is used for storing chat history. Azure OpenAI chat completion and Azure OpenAI Embeddings are used for generation of prompts and semantic search to query a knowledge base for relevant documents.AzureOpenAI
, App Insights
, and Azure blob storage
are used for OpenAI API calls, monitoring and storing chat history.Run the teamsapp deploy
command to trigger deploy stage.
teamsapp deploy --env dev
Run the teamsapp publish
command to trigger publish stage.
teamsapp publish --env dev
Once Teams Copilot Starter app has been published, the app becomes accessible within Microsoft Teams admin center under Teams app
-> Manage app
. From here, Teams administrators can review and approve the app for enterprise users.