Course overview

Course AB-620: Design and build integrated AI agent solutions in Copilot Studio is an advanced, 3-day training program where participants learn how to design, build, and manage production-ready AI agent solutions. The focus is on creating sophisticated agents that integrate seamlessly across systems and operate reliably at an enterprise scale.

Moving beyond standalone agents, this course teaches participants how to deliver end-to-end solutions that automate real business processes, connect securely to enterprise data, and take meaningful action across platforms. Throughout the course, participants will apply best practices for system integration, governance, security, and responsible AI, ensuring their agents are robust and ready for real-world enterprise deployment.

Program highlights

• Planning and designing integrated AI agent solutions for enterprise environments.
• Designing agent conversations, flows, and advanced responses using custom prompts.
• Connecting agents to enterprise knowledge sources via Azure AI Search and Copilot connectors.
• Extending agents using REST APIs, custom connectors, and the Model Context Protocol (MCP).
• Building and orchestrating multi-agent collaboration solutions using the Agent2Agent (A2A) protocol.
• Automating complex scenarios, including human-in-the-loop and computer-use workflows.
• Applying robust security, governance, identity, and responsible AI strategies.
• Implementing Application Lifecycle Management (ALM) to test, monitor, and manage agents in production.

Audience profile

This course is specifically designed for:

• Advanced app builders and professional developers who want to design, extend, and integrate Copilot Studio agents.
• Solution architects designing AI-powered automation for enterprise systems.
• IT professionals responsible for deploying scalable and governed AI solutions.
• Participants aiming to automate business processes by connecting agents to complex enterprise data, systems, and backend services.

Course objectives

After completing this course, participants will be able to:

• Design Copilot Studio agents that integrate deeply with enterprise systems, APIs, and data.
• Build, orchestrate, and manage advanced workflows and multi-agent solutions.
• Automate complex business scenarios with high reliability.
• Apply identity management, security, and responsible AI considerations during the implementation phase.
• Rigorously test, evaluate, monitor, and manage AI agents as true enterprise-grade solutions.

Course details

Module 1: Plan and configure AI agent solutions

• Plan integration with enterprise systems and APIs
• Define identity, security, and governance strategies
• Plan for responsible AI and reusable agent components
• Design agents for internal vs. external audiences

Module 2: Design agent conversations and flows in Copilot Studio

• Create advanced agent flows and human-in-the-loop scenarios
• Configure advanced agent responses with custom prompts and formatting
• Implement error handling and manage input/output parameters
• Add tools, connectors, and variables to topics

Module 3: Connect to enterprise knowledge sources

• Connect agents to Microsoft Power Platform and Copilot connectors
• Integrate enterprise data using Azure AI Search
• Configure generative answers with external data sources

Module 4: Integrate agents with APIs and specialized tools

• Add REST APIs to an agent for backend actions
• Integrate tools using existing custom connectors
• Configure and use Model Context Protocol (MCP) tools
• Automate tasks with computer-use functionalities

Module 5: Configure multi-agent collaboration

• Design multi-agent solutions within Copilot Studio
• Create solutions using the Agent2Agent (A2A) protocol
• Integrate Azure AI Foundry agents and Microsoft Fabric data agents

Module 6: Evaluate, monitor, and manage agents (ALM)

• Evaluate agent performance using test sets and testing methods
• Monitor agent workflows using Application Insights
• Create and manage solutions and environment variables
• Implement Application Lifecycle Management (ALM) using Power Platform Pipelines

To successfully complete the AB-620 course, participants must have solid foundational and intermediate knowledge.

Basic (required) knowledge:

• Generative AI Concepts: Intermediate understanding of AI concepts, including models, orchestration, retrieval-augmented generation (RAG), and prompt engineering.
• API & Integration: Proficiency working with REST APIs, JSON, and common system integration patterns.
• Copilot Studio Experience: Hands-on experience building foundational agents and workflows in Microsoft Copilot Studio.

Recommended prior courses:

To ensure you are fully prepared for this advanced 3-day deep dive, we strongly recommend completing the following courses first:

• PL-7008: Create agents in Microsoft Copilot Studio: Crucial for understanding the core mechanics of building topics, entities, and basic conversational agents.
• MS-4022: Extend Microsoft 365 Copilot in Copilot Studio: Highly recommended to understand how custom knowledge and declarative actions work before advancing to enterprise integrations and multi-agent protocols.

This course serves as the official preparation for Exam AB-620: Designing and Building Integrated AI Solutions in Copilot Studio. Successfully passing this exam validates your skills and awards you the Microsoft Certified: AI Agent Builder Associate certification, demonstrating your ability to design, build, and maintain production-ready AI agents at an enterprise scale.

Exam Details & Registration:

• Exam Link: Exam AB-620: Designing and Building Integrated AI Solutions in Copilot Studio
• Certification Link: Microsoft Certified: AI Agent Builder Associate

Tomaž Vodušek​

Power Platform, Copilot Consultant & Trainer

He began his IT career over 20 years ago as a support engineer at Microsoft's support service, where he supported server and desktop applications. The knowledge he gained in practice was passed on in an educational center, initially in the field of Windows Server Technology, and later he focused on the Microsoft CRM platform.

For the last 17 years, he has been working with the Microsoft CRM business platform, and since 2018, he has been involved with the Power Platform, participating in many projects in Slovenia and the United Kingdom.

He also shares his practical knowledge and experience with students in the field of Microsoft Power Platform through PL courses.

Specialized areas:

• Copilot AI
• Copilot Studio
• Power Platform expert
• Developing on Dataverse
• Power Apps
• Power BI