Bicep template for Content Understanding Studio

Wouter Schaekers 60 Reputation points
2026-05-29T19:15:41.5266667+00:00

Recently, we've deployed foundry resources through bicep templates, including a foundry project for content understanding.
Now with the recent changes, we have to migrate from the preview to GA.

It seems that we don't have to create a foundry project anymore, since the project itself is created in Content Understanding Studio. So for now, we deploy our foundry resource and storage account through bicep, but manually add a project in Content Understanding Studio.

Is there a way to do this through bicep templates? If not, will this be possible in the future?

Azure Content Understanding in Foundry Tools
0 comments No comments

1 answer

Sort by: Most helpful
  1. Jerald Felix 15,690 Reputation points Volunteer Moderator
    2026-05-30T03:11:03.2+00:00

    Hello Wouter Schaekers,

    Greetings! Thanks for raising this question in Q&A forum.

    Great question and the good news is that yes, you can deploy Content Understanding projects through Bicep! The reason it may not have been obvious is that with the GA release, the project resource type has moved under Microsoft.CognitiveServices/accounts/projects, which is different from the old preview approach of using AI Foundry project resources.

    Here's a step-by-step guide to automate your Content Understanding project deployment via Bicep:

    1. Understand the new GA resource structure

    With the GA release (API version 2025-11-01 is now the GA version for Content Understanding), projects in Content Understanding Studio are now represented as Microsoft.CognitiveServices/accounts/projects. This is the resource type you need to deploy via Bicep no separate AI Foundry project is required anymore.

    2. Deploy the Cognitive Services account (AI Foundry resource)

    Make sure your Bicep already deploys the parent Microsoft.CognitiveServices/accounts resource. This is your foundry resource and the parent for the project.

    resource cognitiveServicesAccount 'Microsoft.CognitiveServices/accounts@2025-06-01' = {
      name: 'your-foundry-resource-name'
      location: resourceGroup().location
      kind: 'AIServices'
      sku: {
        name: 'S0'
      }
      properties: {
        publicNetworkAccess: 'Enabled'
      }
    }
    

    3. Add the Content Understanding project as a child resource

    You can now deploy a Microsoft.CognitiveServices/accounts/projects resource directly in Bicep using the 2025-06-01 API version, like this:

    resource contentUnderstandingProject 'Microsoft.CognitiveServices/accounts/projects@2025-06-01' = {
      parent: cognitiveServicesAccount
      name: 'your-project-name'
      location: resourceGroup().location
      properties: {
        description: 'My Content Understanding Project'
        displayName: 'My CU Project'
      }
      tags: {
        environment: 'production'
      }
    }
    

    4. Deploy your Storage Account as before

    Your existing storage account Bicep module stays exactly the same no changes needed there. You just need to ensure the storage account is linked to the Cognitive Services resource if required by your setup.

    5. Deploy the full Bicep template

    Run your deployment as usual:

    az deployment group create \
      --resource-group <your-rg> \
      --template-file main.bicep
    

    6. Verify in Content Understanding Studio

    After deployment, open Content Understanding Studio at https://ai.azure.com and navigate to your resource. The project you deployed via Bicep should appear there and be ready to use just as if you had created it manually through the Studio UI.

    Also worth noting the preview API versions 2025-05-01-preview and 2024-12-01-preview will be retired by July 15, 2026, so migrating to the GA 2025-11-01 API is important to do soon.

    This means you no longer need to create projects manually in the Studio UI at all the full setup from foundry resource to project can now be handled entirely through your Bicep templates in an automated, repeatable way!

    If this answer helps you kindly accept the answer which will help others who have similar questions.

    Best Regards,

    Jerald Felix.

    Was this answer helpful?

    0 comments No comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.