What is Microsoft Copilot for Azure?

If you haven’t tried Microsoft’s new Copilot for Azure yet, I’ve got to tell you—it’s amazing! In this article, I’ll walk you through how Copilot for Azure works, how it can help you reduce costs, and how it’s going to change the way you use Azure daily.

What is Microsoft Copilot for Azure?

Before diving into Microsoft Copilot for Azure, it’s important to understand what Copilot is. The concept starts with Large Language Models (LLMs). These AI models are designed to understand and generate human-like text. They are trained on massive datasets using deep learning algorithms to recognize patterns and provide contextually relevant responses. Essentially, LLMs have access to vast amounts of information, but that doesn’t necessarily mean they fully understand what they are saying.

The breakthrough came with OpenAI's ChatGPT, which popularized LLMs, and now Copilot is built on GPT-4, the latest version of this technology as of now.

Why Bring Copilot into Azure?

If GPT-4 is already out there, why integrate Copilot into Azure? The answer is simple: context.

When you log into the Azure portal, Copilot knows where you are within the system. This means it can provide suggestions and execute tasks based on your role and permissions (RBAC). Copilot not only helps automate repetitive tasks but can also give you real-time insights into security and cost-saving opportunities.

How Copilot for Azure Works

Initial Setup

When you first access Copilot, you may need admin approval to use it. If your admin hasn’t enabled it, you won’t see the tool in action. Once enabled, Copilot guides you through a quick introduction.

Even though Copilot holds a vast amount of knowledge, always remember that it’s essential to verify the information it provides. AI systems can make mistakes, and it’s your responsibility to double-check important outputs.

Security and Permissions

Copilot operates within your user context, which means it can only perform actions that you are authorized to do under Azure’s RBAC system. This feature ensures that Copilot remains secure and only executes actions you have permission for.

However, in its preview version, Copilot has a few limitations:

• You can interact with up to 10 resources at once.
• You are limited to 15 queries per chat and a total of 10 chats within 8 hours.

Real-Time Queries

Once you’re set up, Copilot can begin answering your questions. For example, a simple query like, “How many VMs do I have, and how many are deallocated?” shows Copilot’s power in analyzing your Resource Graph and responding almost instantly. This tool essentially writes and runs the necessary queries for you.

Practical Uses of Copilot in Azure

Virtual Machines

Copilot can easily analyze and recommend cost-saving actions. For example, if you ask, “Is my VM underutilized or overutilized?” Copilot will retrieve metrics from the Resource Graph and offer insights into resizing opportunities. While it may not always execute tasks for you, Copilot provides step-by-step guides or Azure CLI scripts to make changes easier.

Security and Configuration

You can also ask security-related questions in plain language. A simple query like, “Do I have any storage accounts that allow internet traffic?” will generate a list of relevant accounts and their configurations. Copilot helps keep you informed about potential vulnerabilities and security recommendations.

Deploying New Resources with Copilot

Besides monitoring, Copilot can assist in deploying new resources. For instance, if you ask it to help deploy a new VM, Copilot will walk you through the process, ensuring that you have the necessary permissions and resources. You simply click through the prompts, and the Azure CLI script will be executed for you.

Advanced Prompt Engineering

One of the key skills you’ll need to develop when using Copilot is prompt engineering. Writing detailed, clear prompts will help the AI provide better, more accurate responses. For example, instead of just saying, “Check performance,” try, “Check the performance of this Azure VM over the last 24 hours.” The more specific your requests, the better results you’ll get.

Best Practices for Prompts:

• Use terms like fetch, list, retrieve for information gathering.
• Use action verbs like generate, deploy, stop for task execution.
• Be clear about your expertise level to get appropriate advice.

Conclusion

Microsoft’s Copilot for Azure is more than just a helpful AI; it’s a game-changer for admins looking to save time, reduce costs, and enhance security. Whether you’re deploying new resources, monitoring virtual machines, or just looking to simplify everyday tasks, Copilot is here to make your life easier.

Happy learning and experimenting with prompts—Copilot is just getting started!

?