🚀 Agentic AI Workshop - Semantic Kernel

Building the Agentic Future

Welcome to our intensive, hands-on workshop on building agentic AI applications using Microsoft's Semantic Kernel framework! This workshop will take you from foundational concepts to advanced multi-agent orchestration through a series of progressive labs.

📚 Overview

This workshop explores the cutting-edge world of Agentic AI using Microsoft's Semantic Kernel framework. Through a series of four progressive labs, you'll learn how to build, customize, and orchestrate AI agents to perform complex tasks.

🧠 What is Semantic Kernel?

Semantic Kernel is an open-source SDK that integrates Large Language Models (LLMs) with conventional programming languages. It combines the strengths of:

• 🔄 Natural language semantics: Leveraging the power of LLMs to understand and generate human language
• ⚙️ Traditional programming: Using the precision and reliability of conventional code
• 🧩 Extensible architecture: Building flexible plugin systems for enhanced capabilities

Semantic Kernel serves as the foundation for creating AI applications that can reason, learn, and interact in sophisticated ways.

🤖 What is Agentic AI?

Agentic AI refers to autonomous systems that can:

• 🎯 Set and pursue goals: Define objectives and take steps to achieve them
• 🔄 Adapt to feedback: Learn from success and failure
• 🛠️ Use tools and APIs: Interact with external systems and data sources
• 🤝 Collaborate: Work with humans and other agents to solve problems

This workshop focuses on building these capabilities using Semantic Kernel as the underlying framework.

🗺️ Workshop Structure

This workshop is divided into four progressive labs, each building on the knowledge gained in the previous one:

Lab	Title	Focus Area
Lab 01	💡 Introduction to Semantic Kernel	Core concepts, creating a simple plugin
Lab 02	🧰 Working with Built-in Plugins	Leveraging existing plugins, understanding plugin architecture
Lab 03	📡 Model Context Protocol (MCP)	Advanced communication patterns between models and applications
Lab 04	🎭 Multi-Agent Orchestration	Creating agent teams, managing group interactions

Each lab includes hands-on coding exercises, conceptual explanations, and challenges to deepen your understanding.

🧰 Prerequisites

Before starting this workshop, please ensure you have the following:

Knowledge Requirements

• 💻 Intermediate C# programming experience
• 🌐 Basic understanding of RESTful APIs
• 🤖 Familiarity with AI/ML concepts (helpful but not required)

Required Software

• 📊 Visual Studio Code with C# extensions (or Visual Studio 2022 or latest)
• 🛠️ .NET SDK 9.0 or later
• 📋 Git for version control
• 🔄 Azure CLI (for infrastructure deployment)

Required Accounts

• 📝 GitHub account
• ☁️ Microsoft Azure account with active subscription
• 🔑 OpenAI API access (or Azure OpenAI Service access)

🔧 Setup Instructions

Follow these steps to prepare your environment for the workshop:

1. Clone the Repository

git clone https://github.com/..../AgenticWorkshop-SemanticKernel.git
cd AgenticWorkshop-SemanticKernel

2. Install Required .NET Packages

Each lab contains its own project with specific dependencies. To install dependencies for all labs at once:

dotnet restore AgenticWorkshop-SemanticKernel.sln

3. Configure API Keys

Create a .env file in each project directory with your API keys:

# .env file example

AZURE_OPENAI_ENDPOINT=""
AZURE_OPENAI_KEY=""
GITHUB_PERSONAL_ACCESS_TOKEN=""
GOOGLE_API_KEY=""
GOOGLE_SEARCH_ENGINE_ID=""

⚠️ Security Note: Never commit API keys to your repository. The .gitignore file includes .env to help prevent this.

☁️ Infrastructure Deployment

This workshop requires specific Azure resources. Follow these steps to deploy them:

1. Cloud Resources Required

• 🧠 Azure OpenAI Service: For accessing GPT models
• 🌐 Google Custom Search: For Implementing a Built-in Web Search RAG Agent (Lab 2)

2. Automated Deployment

We've created automated deployment scripts to simplify the process:

cd infra
./deploy-workshop-infra.ps1 -SubscriptionId "your-subscription-id" -ResourceGroup "sk-workshop-rg" -Location "eastus"

This script will:

1. 📝 Log you into Azure (if needed)
2. 🌐 Create the resource group if it doesn't exist
3. 🚀 Deploy all required resources with appropriate configurations
4. 🔑 Set up necessary permissions and networking
5. 📝 Output a configuration file for use in the labs

3. Manual Configuration

After deployment completes, you'll need to:

1. 🔍 Create deployments for your Azure OpenAI models

   • GPT-4 Turbo: Recommended for all labs
   • GPT-3.5 Turbo: Minimum for Labs 1-2

2. 🛠️ Configure the deployed resources as needed for your specific environment

💡 Note: Detailed instructions for each step are provided in the infra/README.md file.

🧪 Labs in Detail

💡 Lab 01: Introduction to Semantic Kernel

In this lab, you'll get started with Semantic Kernel fundamentals:

• ✅ Set up a basic Semantic Kernel project
• ✅ Create your first semantic function
• ✅ Implement a simple "Lights Plugin" to control virtual lights
• ✅ Learn about prompt engineering within the Semantic Kernel context

This lab provides a gentle introduction to the framework and establishes the foundation for more complex applications.

🧰 Lab 02: Working with Built-in Plugins

In this lab, you'll expand your knowledge by working with pre-built plugins:

• ✅ Integrate the Google search plugin
• ✅ Chain multiple plugins together
• ✅ Create a conversational agent that uses external data sources
• ✅ Learn how to debug and trace plugin execution

You'll see how Semantic Kernel's plugin architecture enables powerful compositions of capabilities.

📡 Lab 03: Model Context Protocol (MCP)

This lab introduces the Model Context Protocol (MCP) - an advanced communication mechanism:

• ✅ Create an MCP client and server
• ✅ Build a GitHub integration agent
• ✅ Implement a code-fixing agent that can repair bugs
• ✅ Explore request/response filters for enhanced control

MCP represents a significant step forward in managing complex AI interactions and tool usage.

🎭 Lab 04: Multi-Agent Orchestration

The final lab explores the cutting edge of agentic systems - multiple agents working together:

• ✅ Configure specialized agents with different roles
• ✅ Create agent groups with hierarchical structures
• ✅ Implement approval flows and termination strategies
• ✅ Build a multi-agent system that can solve complex problems

You'll learn how multiple agents can collaborate, debate, and achieve goals that would be difficult for single agents.

📚 Learning Resources

📖 Documentation

• Semantic Kernel Official Documentation
• Semantic Kernel GitHub Repository
• Azure OpenAI Service Documentation
• Prompt Engineering Guide

🎥 Videos

• Introduction to Semantic Kernel
• Building AI Agents for Beginners (Playlist)
• Building AI Applications with Semantic Kernel (Playlist)
• Semantic Kernel Deep Dive Session

📝 Articles

• Semantic Kernel: DevBlog
• Agentic AI: A New Paradigm In Autonomous Artificial Intelligence (Forbes)
• Multi-Agent Systems Best Practices

🌐 Community Resources

• Semantic Kernel Discord Community
• Stack Overflow: Semantic Kernel Tag
• GitHub Discussions

🔧 Troubleshooting

Common Issues

Authentication Problems

• ❓ Issue: Azure OpenAI authentication failures
• 🔨 Solution: Check that your API keys are correctly configured in .env and that your Azure role assignments include "Cognitive Services OpenAI User"

Deployment Errors

• ❓ Issue: Resource deployment fails
• 🔨 Solution: Ensure you have sufficient permissions and quota in your Azure subscription. For OpenAI resources, verify your region availability and quota.

👥 Contributing

Contributions to improve this workshop are welcome! Please consider:

• Submitting issues for bugs or suggestions
• Creating pull requests with improvements
• Sharing your experience and examples

Please review our contributing guidelines before making a contribution.

📄 License

This workshop is licensed under the MIT License - see the LICENSE file for details.

---

Created with ❤️ by Microsoft - Marco Antonio Silva

GitHub • Semantic Kernel • Community