Azure AI Workspace Bicep Templates for Azure AI Studio
This repository contains the necessary Bicep templates to spin up an Azure AI Workspace in the new Azure AI Studio.
The new Azure AI Workspace brings together all the features of Azure OpenAI Service, Azure Machine Learning, and other Azure Cognitive Services into a single place. The advantages of the Azure AI Workspace include:
- Leveraging more open-source models for chat, completions, embeddings, text generation, speech recognition, image classification and more, alongside the OpenAI GPT models.
- Create and manage indexes to customize generative AI responses using your own data.
- Generate content filters over your model endpoints to create responsible AI experiences.
- Evaluate, build, test, and deploy your generative AI solutions using Prompt Flow.
Note: The Azure AI Studio is currently in public preview. This preview is provided without a service-level agreement, and is not recommended for production workloads. Certain features might not be supported or might have constrained capabilities. For more information, see Supplemental Terms of Use for Microsoft Azure Previews.
Getting Started
To deploy the Azure AI Workspace, you will need to:
Prerequisites
- Install PowerShell Core.
- Install the Azure CLI.
Deploy the Azure AI Workspace
The main.bicep template contains all the necessary modules to deploy a complete Azure AI Workspace including:
- Azure Resource Group, to contain all the resources.
- Azure Managed Identity, to provide RBAC to the Azure AI Workspace with other resources.
- Azure Storage Account, to store workspace data, and to provide a place for your data.
- Azure Key Vault, to store secrets and keys.
- Azure Log Analytics Workspace, to store logs and metrics.
- Azure Container Registry, to store model images for deployment.
- Azure AI Services, to provide the Azure AI Workspace with access to Azure OpenAI and other Cognitive Services.
- Azure AI Workspace, to provide an environment for you to build, test, and deploy your AI solutions.
- Azure AI Search (optional), to provide a search index over your data.
Configure the deployment parameters
The main.bicepparam file contains all the necessary parameters to deploy a complete Azure AI Workspace including:
- workloadName, used as the suffix for the resource group name, and generate unique names for each deployed resource.
- location, used to specify the Azure region to deploy the Azure AI Workspace.
- newOrExisting, used to specify whether the deployment is new or existing. This must be set to
existing
after deployment as the Workspace Endpoints cannot be modified using Bicep after deployment and will fail otherwise. - includeCognitiveSearch, used to specify whether to include Azure AI Search in the deployment for indexing data.
Using PowerShell & Azure CLI
To deploy the Azure AI Workspace using PowerShell and the Azure CLI:
- Clone this repository.
- Open a PowerShell terminal at the infra folder.
- Login to Azure using the Azure CLI and get a subscription ID.
az login
$subscriptionId = ((az account list -o json --query "[?isDefault]") | ConvertFrom-Json).id
az account set --subscription $subscriptionId
- Deploy the main.bicep template using the Azure CLI.
az deployment sub create --name 'ai-workspace' --location westeurope --template-file ./main.bicep --parameters ./main.bicepparam