Azure Copilot Gets Agents: The Game-Changer for Cloud Operations

If you've been working in Azure lately, you've probably noticed Azure Copilot becoming more capable with each update. But what Microsoft just unveiled in preview takes things to an entirely different level: Agents in Azure Copilot. This isn't just an incremental improvement—it's a paradigm shift that brings agentic AI directly into your Azure workflows.

What Makes Agents Different?

Traditional AI assistants answer questions and provide guidance. Agents take action. They reason through complex scenarios, orchestrate multi-step workflows, and intelligently surface the right capabilities at the right time. With agent mode enabled in Azure Copilot, you're no longer just getting answers—you're getting a virtual cloud solution architect that can plan, execute, and optimize your Azure environment.

Five Specialized Agent Categories

The preview introduces five specialized agent categories, each designed to handle specific operational domains:

Let's dive into what each of these brings to the table and why this matters for anyone managing Azure workloads.

1. Deployments and Infrastructure: Your Virtual Solution Architect

The deployment agent serves as a virtual cloud solution architect, transforming high-level requirements into production-ready infrastructure. This capability goes far beyond template generation—it's about intelligent infrastructure planning grounded in the Azure Well-Architected Framework.

How It Works

Start with a simple prompt like "I need a scalable web app with a SQL database," and the agent orchestrates the entire planning process.

The agent supports deployment planning for all Azure resource types, from simple virtual machines to complex multi-tier architectures involving compute, networking, storage, identity, monitoring, and orchestration.

Real-World Example Prompts

• "Host a sentiment-analysis LLM with Azure Functions for serverless API endpoints, connect to an Azure SQL Database for logging user interactions, and set up alerting for failed requests."
• "Deploy a microservices workload on AKS where API Gateway routes traffic, integrate with Azure Key Vault for secrets, and roll out canary deployments for new service versions."

GitHub Integration That Actually Makes Sense

Here's where it gets really practical: after Azure Copilot generates your Terraform configurations, you can:

This tight integration with developer workflows means the artifacts Azure Copilot generates can seamlessly flow into your existing CI/CD pipelines.

Current Limitations

The deployment agent focuses on "greenfield" scenarios—deploying brand new infrastructure. It doesn't yet support importing or modifying existing infrastructure, though you can still get guidance for those scenarios. Generated artifacts are currently Terraform-only, and while pipeline guidance is provided, automated CI/CD integration isn't available yet.

2. Observability: Intelligent Alert Investigation

When alerts fire in production, every minute counts. The observability agent transforms alert response from a manual investigation process into an AI-assisted workflow that provides actionable insights fast.

How It Works

The agent integrates directly with Azure Monitor to investigate alerts. When you provide an alert (either by viewing it in the portal or providing the alert ID), the agent orchestrates a complete investigation.

You can watch the agent's reasoning in real-time by selecting "Show activity" during the investigation.

Sample Prompts

When viewing an alert in the portal:

• "Can you help investigate this alert?"
• "Start an investigation for this alert."

With a specific alert ID:

• "Troubleshoot this alert: /subscriptions/.../providers/Microsoft.AlertsManagement/alerts/ALERT_ID"

What to Know

Currently, agent capabilities support Application Insights alerts only, with support for other alert types coming in future updates. The agent investigates and recommends solutions but doesn't yet automatically apply remediation steps—you remain in control of any changes.

3. Optimization: Cost and Carbon Reduction Made Easy

In an era where cloud costs and sustainability matter more than ever, the optimization agent helps you identify opportunities to reduce both spending and carbon emissions while maintaining performance.

Capabilities

Integration Points

You can access optimization recommendations through:

• Direct prompts in Azure Copilot chat
• Operations Center (preview) where top recommendations are surfaced
• Azure Advisor cost recommendations page with "Optimize" buttons next to impacted resources

Example Prompts

• "Show me the top five cost-saving opportunities for subscription aaaa0a0a-bb1b-cc2c-dd3d-eeeeee4e4e4e."
• "Explain the recommendation for ContosoVM1."
• "Is there an alternate recommendation for ContosoVM1?"
• "Generate a PowerShell script to apply the recommended optimizations."
• "Summarize total potential cost and carbon reduction from all active recommendations."

Best Practices

For optimal results, include the subscription GUID or ARM Resource URI in your prompts rather than relying solely on resource names. The agent returns five recommendations by default, with a maximum of ten per query.

4. Resiliency: Proactive Protection Against Failures

The resiliency agent helps ensure your Azure resources can withstand and recover from failures, whether from cyber attacks, data corruption, or datacenter outages.

What It Covers

Supported Resource Types for Scripts

Zone resiliency scripts are currently available for:

• Virtual Machines
• App Services
• Azure Database for PostgreSQL
• Azure Database for MySQL
• SQL Managed Instance
• Azure Cache for Redis
• Azure Firewall

Sample Prompts

• "Configure zone resiliency for this resource."
• "Which resources aren't zone-resilient?"
• "How many backup jobs failed in the last 24 hours?"
• "Which data sources don't have a recovery point within the last 7 days?"
• "Increase the security level of this vault."

Important Note

While the agent provides comprehensive guidance and scripts, some advanced capabilities like multi-user authorization and Azure Site Recovery require manual intervention. The agent guides you through these processes with detailed instructions.

5. Troubleshooting and Support: From Problem to Solution Faster

When things go wrong, the troubleshooting agent helps you diagnose issues quickly and often provides one-click fixes to resolve common problems.

How It Works

Resource Coverage

The agent supports troubleshooting for all Azure resource types, with particularly strong capabilities for:

• Azure Cosmos DB
• Azure Virtual Machines
• Azure Kubernetes Service (AKS)

Example Scenarios

• "Help me troubleshoot why my Cosmos DB Cassandra API is failing."
• "I can't connect to my VM, can you help me troubleshoot?"
• "Investigate the health of my pods."
• "Identify reasons for high CPU or memory usage in my AKS cluster."
• "Create a support request for my problem."

What Makes It Special

The integration with Azure Support is brilliant. When the agent can't resolve an issue directly, it doesn't just tell you to open a ticket—it creates one for you with all diagnostic information already gathered. This dramatically reduces the time Microsoft Support needs to understand your issue and start helping.

Getting Started with Agents

Access to Agents in Azure Copilot is managed at the tenant level. For most tenants, access requests are automatically submitted, but availability is being rolled out gradually.

Enabling Agent Mode

Once your tenant has access:

1. Open Azure Copilot as you normally would
2. Look for the agent mode icon in the chat window
3. Select it to enable agent capabilities for the current conversation

With agent mode enabled, Azure Copilot intelligently routes your requests to the appropriate agent. You'll be asked to approve any actions before they're performed—agents won't make changes without your explicit permission.

Current Limitations

• Full support is provided in English only, with limited support for other languages
• Not all functionality is available for every resource type
• Some capabilities are still preview-only for specific scenarios

Want earlier access? Microsoft has a feedback program you can join to help prioritize your organization, though submission doesn't guarantee access.

Why This Matters

We've had AI assistants in development tools for a while now, but Agents in Azure Copilot represents something fundamentally different. This isn't a bolt-on AI feature—it's deeply integrated into the Azure platform with full context awareness, native governance, and the ability to take action on your behalf.

Instead of juggling multiple disconnected tools and manual processes, you get a unified, AI-powered interface that brings together agents, context, and governance in one place.

Looking Forward

This is just the preview. As agent capabilities mature and expand to more resource types and scenarios, the way we manage cloud infrastructure will fundamentally change. The shift from imperative commands to declarative intent—from "do this specific thing" to "achieve this outcome"—is already happening.

The age of agentic cloud management has arrived. The question isn't whether to adopt it—it's how quickly you can get your team up to speed.

Resources:

• Agents (preview) in Azure Copilot
• Deployment capabilities documentation
• Observability capabilities documentation
• Optimization capabilities documentation
• Resiliency capabilities documentation
• Troubleshooting capabilities documentation
• Join the feedback program