Azure  

Getting Started with Azure OpenAI for Power Platform

Introduction

Hey friends!

These days, everyone is talking about AI, GPT, and copilots. As a developer who mostly works with Power Platform, SharePoint, and Power Automate, I was curious: “How do I actually use Azure OpenAI GPT models in my projects?”

So I decided to try it myself and thought, why not share it with you all — step-by-step — like how we talk in the office pantry during a coffee break.

If you're looking to try out GPT models for text generation, summarization, or even chatbot logic, this article will guide you from setting up to testing in Postman, Python, and even Power Automate — with no-nonsense steps.

Let’s get our hands dirty!

What is Azure OpenAI Service?

Azure OpenAI Service gives us access to OpenAI’s powerful models like GPT-3.5, GPT-4, and GPT-4o — directly from Azure with all the enterprise security and scaling features.

These models are capable of,

  • Text generation
  • Email or content writing
  • Summarization
  • Question answering
  • Chatbot conversations

No ML background needed! Just smart prompts, and boom — it works.

Step 1. Create an Azure OpenAI Resource.

  1. Go to https://portal.azure.com
    Azure OpenAI Resource
  2. Search: Azure OpenAI
    Azure OpenAI
  3. Click Create
  4. Fill details
    • Resource group: AzureopenAI
    • Region: East US (only selected regions supported)
    • Name: Azure-openai-rg-demo
    • Pricing tier: Keep it S0
      Pricing tier
  5. Click Create and wait for a few minutes.
    Create

Step 2. Deploy a GPT Model.

Once your resource is ready.

  1. Go to the Azure OpenAI resource.
    Search
  2. Click Go to Azure AI Foundry portal.
    AI Foundry portal
  3. Deployments → + Create
    Deployments
  4. Choose Model: gpt-35-turbo
    Choose Model
  5. Deployment name: mygptchat
    Deployment name
  6. Deploy it.
    Deploy it

That’s it! Now your GPT model is ready to use.

Note. If you already have multiple resource groups in your Azure subscription, double-check where the OpenAI resource is being deployed. Sometimes, if you're using the free tier or limited quota, Azure might auto-select the default resource group or region that has available capacity. It’s always a good idea to review the deployment details before hitting "Create".

Step 3. Try in Azure OpenAI Studio (Playground)

Go to: https://ai.azure.com/resource/playground?wsid=/subscriptions/66225a18-1341-4752-914f-f6652f89ede9/resourceGroups/azureopenai/providers/Microsoft.CognitiveServices/accounts/gkmdo-mbz95rlc-eastus2&tid=c3d16b1b-352f-420f-b679-00e15a434ca6&deploymentId=/subscriptions/66225a18-1341-4752-914f-f6652f89ede9/resourceGroups/azureopenai/providers/Microsoft.CognitiveServices/accounts/gkmdo-mbz95rlc-eastus2/deployments/mygptchat

  1. Select your deployment (mygptchat)
    Select your deployment
  2. Choose Chat playground
  3. Enter a prompt like,

Write a professional apology email for a delayed delivery.

Email

The model will reply with a well-written email. Try different styles: casual, formal, friendly — all using the same prompt but with small tweaks!

Chat History

Step 4. Test GPT Using Postman.

For those who love APIs and testing manually (like me), here’s how you can call GPT using Postman.

Get Your Keys

  • Go to your Azure OpenAI resource → Keys and Endpoints
  • Copy your API key and endpoint
    Test GPT

Postman Setup

Body (raw, JSON)

{

  "messages": [

    { "role": "system", "content": "You are a helpful assistant." },

    { "role": "user", "content": "Summarize this text: Azure OpenAI helps developers access GPT models via REST APIs." }

  ],

  "temperature": 0.7,

  "max_tokens": 200

}

Click Send, and boom — you get your GPT response.

GPT Response

Step 5. Try It in Python.

If you prefer coding, here’s a Python example using requests.

python

url = "https://gkmdo-mbz95rlc-eastus2.cognitiveservices.azure.com/openai/deployments/gpt4Demo/chat/completions?api-version=2025-01-01-preview"

headers = {

    "api-key": "YOUR_API_KEY",

    "Content-Type": "application/json"

}

data = {

    "messages": [

        {"role": "system", "content": "You are a helpful assistant."},

        {"role": "user", "content": "Write a tweet about Azure OpenAI"}

    ],

    "temperature": 0.5,

    "max_tokens": 100

}
response = requests.post(
    url,
    headers=headers,
    json=data
)

print(
    response.json()["choices"][0]["message"]["content"]
)

You’ll see your tweet instantly!

Bonus: Connect GPT with Power Automate

Yes, you can call the same GPT endpoint in Power Automate using the HTTP action!

  1. Use HTTP → POST
  2. Headers
    • api-key: your Azure key
    • Content-Type: application/json
  3. Body
{

  "messages": [

    {"role": "system", "content": "You are an email generator."},

    {"role": "user", "content": "Write an email asking for project update"}

  ],

  "max_tokens": 150

}

Use the response to send email, update SharePoint, etc.

Super powerful when combined with Power Apps or Copilot Studio!

Final Thoughts

Learning Azure OpenAI is not just for AI engineers — as a Power Platform or full-stack developer, you can build smart, context-aware solutions that save time and impress clients.

  • Start small with Postman
  • Build something in Python
  • Automate it with Power Automate
  • And keep exploring use cases like resume matcher

Membership not found