Skip to main content

Deploying an Azure OpenAI Model and Connecting It to Progress Agentic RAG

Summary Flow​

  1. Create Azure account and subscription
  2. Create Azure OpenAI resource
  3. Open Microsoft Foundry from the resource overview
  4. Deploy a model via the Model Catalog
  5. Retrieve API key and endpoint from Foundry UI
  6. Configure Azure OpenAI in Progress Agentic RAG

1. Azure Account, Subscription, and Access (Prerequisite)​

Before deploying a model, you must have:

  • An active Azure account
  • An Azure subscription with billing enabled
  • Access to Azure OpenAI Service

Depending on your tenant and region, Azure OpenAI access may require approval.

Outcome

  • A subscription capable of hosting Azure OpenAI resources

2. Create an Azure OpenAI Resource​

This step is completed in the Azure Portal.

Steps​

  1. Create a new Azure OpenAI resource
  2. Specify:
    • Subscription
    • Resource Group
    • Region
    • Pricing tier Microsoft Azure Open AI Configuration
  3. Complete resource creation

The pricing tier determines billing and usage limits and must be selected at creation time.

Outcome

  • A named Azure OpenAI resource that will host your model deployments
  • This resource will later supply the endpoint URL

3. Access Microsoft Foundry from the Azure OpenAI Resource​

Once the Azure OpenAI resource is created, model deployment and configuration occur in Microsoft Foundry.

Steps​

  1. Open the Azure OpenAI resource in the Azure Portal
  2. From the Overview page:
    • Scroll to Explore and deploy
    • Select the option to open the Foundry experience

This routes you to the Foundry instance associated with your specific Azure OpenAI resource.

Outcome

  • You are now operating inside Foundry, scoped to your Azure OpenAI resource

4. Deploy an LLM Using the Foundry Model Catalog​

All model deployment configuration happens in Foundry.

Steps​

  1. Open the Model Catalog from the left-hand navigation Model Catalog in Microsoft Foundry UI
  2. Select the desired model (for example: gpt-4o-mini)
  3. Configure the deployment:
    • Deployment name
    • Deployment type
    • Model version
    • Model upgrade policy
    • Any additional deployment options Model Deployment in Microsoft Foundry UI
  4. Deploy the model

Verification​

  • Open the Deployments tab in the left-hand navigation
  • Confirm the deployment appears and is active

Important distinctions

  • Model name: the actual model selected (e.g. gpt-4o-mini)
  • Deployment name: a custom identifier you define and must reference exactly later

Outcome

  • A live Azure OpenAI model deployment ready to receive requests

5. Retrieve API Key and Endpoint URL from Foundry​

Authentication details are retrieved directly from Foundry.

Steps​

  1. Navigate to the Home page in Foundry
  2. Copy:
    • API key
    • Azure OpenAI endpoint URL

Outcome

  • Credentials required to authenticate external clients

6. Configure Azure OpenAI in Progress Agentic RAG​

This step connects your deployed model to Progress Agentic RAG.

Required Configuration Values​

FieldSource
API KeyFoundry
Endpoint URLFoundry
Deployment NameFoundry
Model NameFoundry

Steps​

  1. Select Azure OpenAI as the LLM provider
  2. Enable Use your own Azure OpenAI key
  3. Enter:
    • API Key
    • Endpoint URL
    • Deployment Name
    • Model Name
  4. Save the configuration
  5. Test answer generation

Outcome

  • Progress Agentic RAG uses your Azure-hosted LLM for answer generation