Skip to content

Provision your own Azure Foundry deployment ​

Follow these steps if you want to run the labs against your own Azure subscription instead of the pre-deployed shared resources.

Skip this if you use pre-deployed resources

If you are happy to use the shared workshop resources, go to Pre-deployed resources instead and copy the values straight into your .env file.

Prerequisites ​

  • An Azure subscription with the ability to create resources (Contributor or Owner role on a resource group).
  • The Azure CLI installed and logged in (az login).

Step 1 — Create a resource group ​

bash
az group create \
  --name rg-foundry-workshop \
  --location eastus

Pick any region that has Azure OpenAI quota available. eastus and swedencentral tend to have the most capacity.


Step 2 — Create an Azure OpenAI resource ​

bash
az cognitiveservices account create \
  --name aoai-workshop \
  --resource-group rg-foundry-workshop \
  --kind OpenAI \
  --sku S0 \
  --location eastus \
  --yes

Wait for the command to complete (roughly 30–60 seconds).


Step 3 — Deploy a model ​

bash
az cognitiveservices account deployment create \
  --name aoai-workshop \
  --resource-group rg-foundry-workshop \
  --deployment-name gpt-5.5 \
  --model-name gpt-5.5 \
  --model-format OpenAI \
  --sku-capacity 10 \
  --sku-name "Standard"

Model availability

If the command asks for a model version, list the versions available in your region with az cognitiveservices model list --location eastus -o table and pass it via --model-version. If gpt-5.5 is not available in your chosen region, substitute another chat model such as gpt-4o. Update AZURE_OPENAI_DEPLOYMENT in your .env accordingly.


Step 4 — Retrieve your endpoint and key ​

Endpoint:

bash
az cognitiveservices account show \
  --name aoai-workshop \
  --resource-group rg-foundry-workshop \
  --query "properties.endpoint" -o tsv

API key:

bash
az cognitiveservices account keys list \
  --name aoai-workshop \
  --resource-group rg-foundry-workshop \
  --query "key1" -o tsv

Copy both values — you will need them in the next step.

Alternatively, open the Azure Portal, navigate to your Azure OpenAI resource, and find the values under Resource Management → Keys and Endpoint.


Step 5 — Fill in your .env file ​

In the root of the workshop repo, copy the example file and fill in your values:

bash
copy .env.example .env
bash
cp .env.example .env

Then open .env and set:

dotenv
AZURE_OPENAI_ENDPOINT=<your-endpoint>       # e.g. https://aoai-workshop.openai.azure.com/
AZURE_OPENAI_KEY=<your-key>
AZURE_OPENAI_DEPLOYMENT=gpt-5.5              # or whichever model name you deployed

Verify the deployment ​

Run the Part 1 baseline agent as a smoke-test (from the root of your dm-ws clone, with the virtual environment active):

powershell
copy baseline\01-1-first-agent.py agent.py
python agent.py --ask "Where is order #1037?"
bash
cp baseline/01-1-first-agent.py agent.py
python agent.py --ask "Where is order #1037?"

Expected output: a short natural-language answer about the order — that confirms your endpoint, key, and deployment name all work.

If you see an authentication error, double-check the endpoint trailing slash and that the key was copied without extra whitespace.