Create Azure OpenAI Service
To use the Azure OpenAI Service, you need to create the required service in your own Azure resource. This is necessary to ensure the security of your data and to keep costs fully transparent and under your control.
The setup is done outside of Business Central. Since navigation in the Azure Portal can quickly become complex, this guide will walk you through the process step by step.
Prerequisites
What you need
- A Microsoft Azure account with administrator rights
- An active Azure subscription that is already set up in your Azure account
- The required Azure resource providers for the AI services being used, which are registered in the Azure subscription. Instructions can be found at Register Azure Resource Providers.
Part 1: Create Azure OpenAI Resource
Step 1: Open Azure Portal
Navigate to the Azure OpenAI section in the Azure Portal: Azure OpenAI in Azure Portal
Step 2: Create New Resource
Click on “Create" and select Azure OpenAI from the list.
Official Microsoft Documentation
Microsoft provides comprehensive documentation for this creation process. If you are unsure about certain options, you can find more information here: Create Azure OpenAI Resource
Go through the wizard and complete the resource creation.
Part 2: Set Up Microsoft Foundry
To connect your Azure OpenAI resource with Business Central, you need a Microsoft Foundry project.
No Foundry project yet? Expand this section
Create Foundry Project
- Navigate to the Microsoft AI Foundry Portal
Disable New Foundry Switch
Make sure the “New Foundry" switch is set to Off. Not all required features are available in the new Foundry yet.
- Click on “Create New"
- Select the recommended option “Microsoft Foundry resource" and click Next
-
Create a new project with a meaningful, company-specific name, for example:
Simova-Foundry
If there is no suitable resource available, create a new one directly in this wizard.
- Click on Create and wait for the deployment to complete.
Step 3: Open Foundry Project
Once the project is ready, navigate into it by clicking on the project name.
Step 4: Open Management Center
Click on “Management center" in the lower left corner on the left side.
Step 5: Add New Connection
- Navigate to Connected Resources
- Click on “New Connection"
- Select Azure OpenAI from the connection type list
Step 6: Add Resource
Your newly created Azure OpenAI resource will be displayed in the list. Click on Add to connect it.
Step 7: Verify Connection
Your resource should now be listed under Connected Resources. The Azure OpenAI resource is successfully linked to your Foundry project.
Step 8: Retrieve Credentials
Navigate to your credentials as follows:
- Open the Management center (lower left corner of your Foundry project)
- Navigate to Connected Resources
- Click on your newly added Azure OpenAI resource to open it
You will find the values you need to set up the Business Central AI service:
| Field | Description |
|---|---|
| Target | The endpoint URL of the service |
| API Key | The authentication key |
Part 3: Deploy AI Model
Before you can use the service, you need to deploy an AI model.
Step 9: Navigate to Models & Endpoints
Click on “Models + endpoints" in the left navigation of the Foundry hub.
Step 10: Deploy Model
Click on “Deploy Model", select the desired model for the CORE AI service, and click Deploy.
Model Recommendation
Always select the latest available model version for the best performance and accuracy.
Attention
When manually deploying an AI model, select a standard deployment type such as Global Standard or Data Zone Standard.
Provisioned Throughput should only be used when permanently reserved capacity is needed. This deployment type incurs ongoing costs even if the model is not used.
Also, note that Azure Cost Management Budgets only send notifications and do not automatically stop costs.
Final Configuration in Business Central
Once the model is deployed, enter the following three values into your Business Central AI Service configuration:
| BC Field | Source |
|---|---|
| Target Endpoint | From the connected resource → Target |
| API Key | From the connected resource → API Key |
| Model Name | The name of the model just deployed |
That's it – you are ready to go!
Important: Validation Limitation
The validation of the Business Central AI Service cannot determine if your resource is correctly configured – it can only check if the resource exists.
Test your entire process from start to finish after completing the setup to ensure everything works as expected.