Architecting Intelligent Agents in Azure: Building Agentic Systems with Python and the Microsoft Agent Framework

Architecting Intelligent Agents in Azure takes you from your first interaction with an AI agent to deploying a fully orchestrated, production-ready system that can reason, remember, collaborate, and act.

Using the Microsoft Agent Framework and Python, this book walks you through every engineering layer behind modern agentic systems, from grounding and memory to tools, semantic recall, and autonomous action.

Across the chapters, you'll build Thain, an Azure-native feedback agent that evolves into a multi-agent system with the ability to retrieve knowledge, coordinate tasks, and refine its performance.

With a balanced mix of architecture, hands-on code, and cloud patterns, you'll learn how Azure Cosmos DB, Azure AI Search, Azure Monitor, and serverless components come together to form intelligent, self-improving enterprise agents.

Whether you're an AI engineer, Azure developer, or solution architect, this book offers a practical, end-to-end guide to building agents that can reason, recall, collaborate, and grow over time.

What You Will Learn:

Build a fully functioning agent using Python and the Microsoft Agent Framework Implement reasoning loops, short-term memory, and tool-based actions Add persistent memory using Azure Cosmos DB and semantic recall using Azure AI Search Create safe and governed agents with telemetry, observability, and policy enforcement Integrate external systems through tools for tickets, documents, notifications, and workflows Orchestrate collaborative multi-agent systems with shared memory Deploy agentic workloads using Azure Functions, CI/CD pipelines, and cloud-native architecture Optimize cost, scale, and performance for enterprise production environments

Who This Book Is For:

AI engineers and developers building real agentic systems and evaluating the Microsoft Agent Framework, vector search, and Azure AI services

Cloud and Azure engineers looking to integrate AI capabilities deeply into existing or new applications

Solution architects designing AI-native or AI-augmented enterprise platforms

Full-stack engineers transitioning into AI engineering and wanting an end-to-end practical pathway into agentic systems