1. Select a Template

!!! tip "BY THE END OF THIS MODULE YOU WILL BE ABLE TO"

- [ ] Describe what AZD templates are
- [ ] Discover and use AZD templates for AI
- [ ] Get started with the AI Agents template
- [ ] **Lab 1:** AZD Quickstart in Codespaces or a dev container

---

1. A Builder Analogy

Building a modern enterprise-ready AI application from scratch can be daunting. It's a little bit like building your new home on your own, brick by brick. Yes, it can be done! But it is not the most effective way to get the desired end result!

Instead, we often start with an existing design blueprint, and work with an architect to customize it to our personal requirements. And that's exactly the approachto take when building intelligent applications. First, find a good design architecture that fits your problem space. Then work with a solution architect to customize and develop the solution for your specific scenario.

But where can we find these design blueprints? And how do we find an architect that is willing to teach us how to customize and deploy these blueprints on our own? In this workshop, we answer those questions by introducing you to three technologies:

1. Azure Developer CLI - an open-source tool that accelerates the developer path in going from local development (build) to cloud deployment (ship).
2. Microsoft Foundry Templates - standardized open-source repositories containing sample code, infrastructure and configuration files for deploying an AI solution architecture.
3. GitHub Copilot Agent Mode - a coding agent grounded in Azure knowledge, that can guide us in navigating the codebase and making changes - using natural language.

With these tools in hand, we can now discover the right template, deploy it to validate it works, and customize it to suit our specific scenarios. Let's dive in and learn how these work.

---

2. Azure Developer CLI

The Azure Developer CLI (or azd) is an open-source commandline tool that can speed up your code-to-cloud journey with a set of developer-friendly commands that work consistently across your IDE (development) and CI/CD (devops) environments.

With azd, your deployment journey can be as simple as:

• azd init - Initializes a new AI project from an existing AZD template.
• azd up - Provisions infrastructure and deploy your application in one step.
• azd monitor - Get real-time monitoring and diagnostics for your deployed application.
• azd pipeline config - Setup CI/CD pipelines to automate deployment to Azure.

🎯 | EXERCISE:
Explore the azd commandline tool in your current workshop environment now. This can be GitHub Codespaces, a dev container, or a local clone with the prerequisites installed. Start by typing this command to see what the tool can do:

azd help

---

3. The AZD Template

For azd to achieve this, it needs to know the infrastructure to provision, the configuration settings to enforce, and the application to deploy. This is where AZD templates come in.

AZD templates are open-source repositories that combine sample code with infrastructure and configuraton files required for deploying the solution architecture. By using an Infrastructure-as-Code (IaC) approach, they allow template resource definitions and configuration settings to be version-controller (just like the app source code) - creating reusable and consistent workflows across users of that project.

When creating or reusing an AZD template for your scenario, consider these questions:

1. What are you building? → Is there a template that has starter code for that scenario?
2. How is your solution architected? → Is there a template that has the necessary resources?
3. How is your solution deployed? → Think azd deploy with pre/post-processing hooks!
4. How can you optimize it further? → Think built-in monitoring and automation pipelines!

🎯 | EXERCISE:
Visit the Awesome AZD gallery and use the filters to explore the 250+ templates currently available. See if you can find on that aligns to your scenario requirements.

---

4. AI App Templates

For AI-powered applications, Microsoft provides specialized templates featuring Microsoft Foundry and Foundry Agents. These templates accelerate your path to building intelligent, production-ready applications.

Microsoft Foundry & Foundry Agents Templates

Select a template below to deploy. Each template is available on Awesome AZD and can be initialized with a single command.

Template	Description	Deploy Command
AI Chat with RAG	Chat application with Retrieval Augmented Generation using Microsoft Foundry	azd init -t azure-samples/azure-search-openai-demo
Foundry Agent Service Starter	Build AI agents with Foundry Agents for autonomous task execution	azd init -t azure-samples/foundry-agent-service-starter
Multi-Agent Orchestration	Coordinate multiple Foundry Agents for complex workflows	azd init -t azure-samples/multi-agent-orchestration
AI Document Intelligence	Extract and analyze documents with Microsoft Foundry models	azd init -t azure-samples/ai-document-processing
Conversational AI Bot	Build intelligent chatbots with Microsoft Foundry integration	azd init -t azure-samples/ai-chat-protocol
AI Image Generation	Generate images using DALL-E via Microsoft Foundry	azd init -t azure-samples/ai-image-generation
Semantic Kernel Agent	AI agents using Semantic Kernel with Foundry Agents	azd init -t azure-samples/semantic-kernel-agent
AutoGen Multi-Agent	Multi-agent systems using AutoGen framework	azd init -t azure-samples/autogen-multi-agent

Quick Start

1. Browse templates: Visit https://azure.github.io/awesome-azd/ and filter by AI, Agents, or Microsoft Foundry
2. Select your template: Choose one that matches your use case
3. Initialize: Run the azd init command for your chosen template
4. Deploy: Run azd up to provision and deploy

🎯 | EXERCISE:
Select one of the templates above based on your scenario:

• Building a chatbot? → Start with AI Chat with RAG or Conversational AI Bot
• Need autonomous agents? → Try Foundry Agent Service Starter or Multi-Agent Orchestration
• Processing documents? → Use AI Document Intelligence
• Want AI coding assistance? → Explore Semantic Kernel Agent or AutoGen Multi-Agent

azd init -t azure-samples/azure-search-openai-demo
azd up

!!! info "Explore More Templates" The Awesome AZD Gallery contains 250+ templates. Use the filters to find templates matching your specific requirements for language, framework, and Azure services.