📚 Course: AZD For Beginners | ⏱️ Duration: 2-3 hours | ⭐ Complexity: Advanced
This chapter covers advanced multi-agent architecture patterns, agent orchestration, and production-ready AI deployments for complex scenarios.
Validated against
azd 1.23.12in March 2026.
By completing this chapter, you will:
- Understand multi-agent architecture patterns
- Deploy coordinated AI agent systems
- Implement agent-to-agent communication
- Build production-ready multi-agent solutions
| # | Lesson | Description | Time |
|---|---|---|---|
| 1 | Retail Multi-Agent Solution | Complete implementation walkthrough | 90 min |
| 2 | Coordination Patterns | Agent orchestration strategies | 30 min |
| 3 | ARM Template Deployment | One-click deployment | 30 min |
# Option 1: Deploy from a template
azd init --template agent-openai-python-prompty
azd up
# Option 2: Deploy from an agent manifest (requires azure.ai.agents extension)
azd extension install azure.ai.agents
azd ai agent init -m agent-manifest.yaml
azd upWhich approach? Use
azd init --templateto start from a working sample. Useazd ai agent initwhen you have your own agent manifest. See the AZD AI CLI reference for full details.
graph TD
Orchestrator[Orchestrator Agent<br/>Routes requests, manages workflow] --> Customer[Customer Agent<br/>User queries, preferences]
Orchestrator --> Inventory[Inventory Agent<br/>Stock levels, orders]
The Retail Multi-Agent Solution demonstrates:
- Customer Agent: Handles user interactions and preferences
- Inventory Agent: Manages stock and order processing
- Orchestrator: Coordinates between agents
- Shared Memory: Cross-agent context management
| Service | Purpose |
|---|---|
| Microsoft Foundry Models | Language understanding |
| Azure AI Search | Product catalog |
| Cosmos DB | Agent state and memory |
| Container Apps | Agent hosting |
| Application Insights | Monitoring |
| Direction | Chapter |
|---|---|
| Previous | Chapter 4: Infrastructure |
| Next | Chapter 6: Pre-Deployment |