� REPOSITORY MODERNIZATION IN PROGRESS | January 2026 🚧

⚠️ IMPORTANT NOTICE: This repository is currently being upgraded to align with the latest Microsoft Foundry architecture and SDK patterns. During this transition period:

• ❌ The repository may not function correctly
• 🔄 Breaking infrastructure changes are being implemented
• 📅 Expected completion: Late February 2026

What's Changing?

• ✨ Modern CognitiveServices architecture (replacing MachineLearningServices Hub/Project)
• 🔐 Keyless authentication with Entra ID (DefaultAzureCredential)
• 🎯 Simplified infrastructure (60% less complexity)
• 🚀 Latest azure-ai-projects SDK patterns

👉 See CHANGELOG.md for detailed migration information

Welcome to the "Build Agents with Microsoft Foundry" workshop repo! 🎉

Dive into the world of intelligent conversational agents with Microsoft Foundry & AI Agent Service, a seamless blend of service and SDK that simplifies the development of robust AI-driven solutions. In this hands-on workshop, you'll learn to create a powerful agent capable of answering sales-related queries, performing data analysis, generating visualizations, and integrating external data sources to deliver enhanced business insights. 🚀

🌟 Beyond Chatbots: Unlock the Future of AI Applications 🌟

💡 Think Bigger, Build Smarter!
Traditional chatbots and Copilots like ChatGPT have revolutionized how we interact with AI. But why stop there? This workshop challenges you to think beyond chatbots and explore the next frontier: Agentic Applications.

🚀 What Are Agentic Applications?

Agentic applications are AI systems with superpowers:

• 🗣️ Conversational Intelligence: Engage users naturally.
• 🤖 Action-Oriented Capabilities: Automate workflows and execute meaningful actions.
• 📊 Proactive Insights: Deliver transformative solutions to complex business challenges.

🔑 Why Go Beyond Chatbots?

By embracing Agentic applications, you’ll:

• Unlock New Possibilities: Tackle advanced use cases that traditional chatbots can’t handle.
• Drive Innovation: Build solutions tailored to your unique business needs.
• Empower Decision-Making: Enable smarter, faster, and more impactful decisions.

🌐 Your Journey Starts Here

This workshop is your gateway to creating intelligent, action-driven AI solutions. Let's redefine what's possible with Microsoft Foundry and take your AI development to the next level! 💼✨

⏱️ Duration

🕒 45 - 60 minutes

🎥 Watch a Recent Workshop Execution Video

Catch a glimpse of the workshop in action! Watch the latest execution of this workshop to see how to build intelligent agents step-by-step. Click the thumbnail below to view the video:

💡 The Use Case for This Lab

Imagine you are a sales manager at Contoso, a multinational retail company that sells outdoor equipment. 🏕️ You need to analyze sales data to find trends, understand customer preferences, and make informed business decisions. To help you, Contoso has developed a conversational agent that can answer questions about your sales data. 💬📈

🎯 What You Will Learn

By the end of this workshop, you will:

• 🛠️ Build an agent app using Azure AI Agent Service.
• 🔍 Explore its tools.
• 📜 Effectively use instructions to guide the LLM.

After the basic workshop we will guide you to continue enhancing your agent with more advanced features, such as:

• 🚀 Expand the agent's capabilities beyond basic conversational tasks.
• 🧠 Integrate advanced AI features such as reasoning, planning, and decision-making.
• 🔗 Connect the agent to external APIs and data sources for enriched functionality.
• 📊 Enable the agent to generate visualizations and actionable insights from datasets.
• 🤝 Incorporate multi-agent collaboration to solve intricate business challenges.
• 🌐 Leverage Azure AI services to scale and enhance the agent's performance.

📖 Workshop Instructions

👉 Start the full workshop here: Introduction

📚 Additional Resources

When you finished the workshop, check out additional resources such as the samples in the samples folder and the official documentation.

Samples overview

Agent Creation:

• Create AI Agents using the AI Foundry UI
• Create AI Agents using Jupyter Notebooks
• Create RAG Agents using Jupyter Notebooks
• Create Multi-Agent Systems using Jupyter Notebooks
• Create Agents with the Bing Grounding Tool

Deep Research Tool:

• Create Agents with the Deep Research Tool (under revision - o3-deep-research deployment temporarily deactivated)
• Debugging sample for the Deep Research Tool (under revision)

Agent Frameworks:

• Create Agents using Semantic Kernel (will be upgraded to Agent Framework samples)
• Create Agents using AutoGen.... (coming soon)
• Create Agents using the Microsoft Agent Framework.... (coming soon)

Autonomous Agents:

• Create Autonomous Agents....(coming soon)

Sovereign AI using Foundry Local:

• Setup Foundry Local to go Sovereign
• Use OpenAI SDK with Foundry Local

Build User Interfaces:

• Create a (mockup) User Interface for Agents using Vibe Coding
• Create a NLWeb UI for Agents....(coming soon)

MCP Integration:

• Create AI Foundry Agents connected to MCP Servers

• Create MCP Server....(coming soon)

Multimodal Agents:

• Create a Talking Avatar website using Azure Text-to-Speech Avatar Real-Time API
• Create Multimodal Agents....(coming soon)
• Create Agents with Vision Tool....(coming soon)
• Create Agents with Speech Tool....(coming soon)

Integration with Business Applications:

• Create Agents with SAP Connector....(coming soon)
• Create Agents with Salesforce Connector....(coming soon)

Compliance, Observability & Management:

• Implement Content Safety and Responsible AI....(coming soon)
• Monitor Agent Performance with Application Insights....(coming soon)
• Set up Logging and Tracing for Agents....(coming soon)
• Implement Cost Management and Budgeting....(coming soon)
• Deploy Agents with Azure DevOps/GitHub Actions....(coming soon)
• Manage Agent Lifecycle and Versioning....(coming soon)
• Implement Compliance Controls (GDPR, HIPAA, etc.)....(coming soon)

Resources for organizing a hackathon 🎉

If you are interested in organizing a hackathon using this repository, check out the infra folder. 📂 It contains scripts and templates to set up the Azure infrastructure needed for the hackathon, as well as policies to control costs. 💰 Below are the links to the guides:

• Hackathon Infrastructure Setup Guide: 🛠️ Learn how to set up the necessary Azure infrastructure for hosting a hackathon, including cost control policies and deployment scripts. Read more here.
• Cost Control Policies Guide: 📊 Understand the cost control policies implemented to manage and monitor expenses during the hackathon. Read more here.

📝 Recent Changes - January 2026

🔄 Major Infrastructure Modernization

Key Changes:

• New Architecture: Migrated from MachineLearningServices Hub/Project to modern CognitiveServices-based Microsoft Foundry
• Simplified Infrastructure: Removed 4 dependent resources (Storage, KeyVault, AppInsights, Container Registry) - now just 2 core resources
• Keyless Authentication: Entra ID (DefaultAzureCredential) is now the default - no more API key management
• Modern Endpoints:
  • Project endpoint: https://{resource}.services.ai.azure.com/api/projects/{project}
  • Inference endpoint: https://{resource}.services.ai.azure.com/models
• SDK Alignment: Ready for azure-ai-projects SDK (Python/JS/.NET)

Impact:

• ⚠️ Breaking change - requires fresh deployment
• ✅ 34% smaller templates (1246 → 823 lines)
• ✅ Best practice security (keyless auth)
• ✅ Aligned with Microsoft Foundry roadmap

Files Updated:

• infra/main.bicep - Complete rewrite with modern architecture
• infra/modules/ai-foundry.bicep - NEW: Microsoft Foundry resource
• infra/modules/ai-foundry-project.bicep - NEW: Project as child resource
• infra/main.parameters.json - Updated parameter structure
• docs/docs/getting-started.md - Updated Azure CLI login commands

Migration Guide: See CHANGELOG.md for detailed migration steps and breaking changes.

Old Architecture Backup: Available in infra/main-old-hub-architecture.bicep