fr
en
Become a Microsoft Certified AI Agent Builder Associate (AB-900/AB-620)
Connect Copilot administration with agent design to support an AI project from end to end. This AB-900/AB-620 pathway helps you control access, connect business systems and manage the agent lifecycle. Expand your ability to deploy useful solutions for teams while preparing for AI Agent Builder Associate certification.
- Duration
- 5 days 35 hours
- Code
- AB-620-BIS Code
- Certification
- Microsoft 365 Certified : AI Agent Builder Associate Certification
Accredited training for the Microsoft 365 Certified : AI Agent Builder Associate certification.
Presentation
Administering a Microsoft 365 Copilot environment is no longer sufficient when an organisation wants to deploy its own tailored AI agents. You need to understand governance and data security fundamentals while also being able to design, integrate and manage agents that connect with enterprise systems.
This 4-day pathway combines both dimensions. You will begin with 1 day covering the fundamentals of Copilot and agent administration, including licensing, governance and data protection with Microsoft Purview. The following 3 days focus on advanced agent design with Copilot Studio: data source integration, multi-agent orchestration, MCP and Agent2Agent (A2A) protocols, testing and production management.
By the end of the pathway, you will be ready to take the AB-620 examination included in our offer. Passing leads to Microsoft Certified: AI Agent Builder Associate certification (see the Certification tab for details).
Objectives
By the end of this course, you will be able to:
- identify the key features and objects of Microsoft 365 services;
- understand data protection and governance principles for Microsoft 365 and Copilot with Microsoft Purview;
- perform basic administrative tasks for Microsoft 365 Copilot and agents, including licensing, lifecycle management and monitoring;
- plan and configure an agent solution for a business scenario;
- integrate an agent with data sources and external systems through connectors, MCP and RAG;
- orchestrate multiple agents using the Agent2Agent (A2A) protocol;
- test, debug and publish an agent to appropriate channels;
- manage an agent's lifecycle, governance and security in production;
- pass AB-620 and earn Microsoft Certified: AI Agent Builder Associate certification.
Program
AB-900: Microsoft 365 Copilot and agent administration fundamentals (1 day)
Module 1: Identify key Microsoft 365 features and objects
- Microsoft 365 objects: users, groups, teams, sites and libraries.
- Licence management and its impact on service access.
- Navigate the Microsoft 365 admin centre.
- Zero Trust security principles, authentication and authorisation.
- Microsoft Defender XDR and threat protection capabilities.
- The role of Privileged Identity Management (PIM) and application registrations.
Lab
- Navigate the Microsoft 365 admin centre and review audit logs.
Module 2: Understand data protection and governance for Microsoft 365 and Copilot
- How Copilot accesses data and how Microsoft Graph influences its responses.
- Protection controls and permissions in Microsoft 365, Purview and Defender.
- Responsible AI principles.
- Microsoft Purview capabilities: data classification, sensitivity labels and data loss prevention (DLP).
- Insider risk management, communication compliance and the data lifecycle.
- Identify oversharing in SharePoint and apply restricted access control.
Lab
- Configure a sensitivity label and DLP policy for a Copilot environment.
Module 3: Perform basic administrative tasks for Copilot and agents
- Manage Microsoft 365 Copilot licences.
- Monitor Copilot use through analytics reports.
- Create and manage Copilot agents.
- The agent lifecycle: creation, approval and monitoring.
Lab
- Create a Copilot agent and review its usage reports.
AB-620: Design integrated AI agents with Copilot Studio (3 days)
Module 4: Plan and configure an agent solution
- Plan an agent architecture suited to a business scenario.
- Configure topics, agent flows and generative answers.
- Use adaptive cards and custom prompts with the Foundry model catalogue.
- Work with Dataverse tables, columns and relationships, and Power Fx expressions.
- Manage environments, solutions and application lifecycle management (ALM), including environment variables and Power Platform pipelines.
Lab
- Plan and configure an initial agent with topics, flows and an adaptive card.
Module 5: Integrate and extend agents in Copilot Studio
- Ground an agent in actual organisational data, including retrieval-augmented generation (RAG) with Azure AI Search.
- Add tools to an agent: custom connectors, REST APIs and Model Context Protocol (MCP) servers.
- Configure computer use so that an agent can act on a user's behalf.
- Orchestrate multiple agents using the Agent2Agent (A2A) protocol.
- Extend an agent with Power Automate to automate complex tasks.
Lab
- Build a RAG-grounded agent, integrate a REST API tool and orchestrate a second agent through A2A.
Module 6: Test and manage agents
- Test agent flows and handle errors.
- Debug and diagnose agent behaviour.
- Publish an agent to appropriate channels, including Teams, websites and applications.
- Monitor, govern and secure an agent in production.
- Manage the agent lifecycle from creation to retirement.
Lab
- Test an agent with error handling, publish it to Teams and configure monitoring.
Audience
This course is designed for professionals wishing to administer and design Copilot and AI agent solutions, particularly:
- Microsoft 365 administrators seeking Copilot administration fundamentals before progressing to agent design;
- professional developers wishing to build AI agents integrated with enterprise systems;
- solution architects and IT consultants supporting enterprise-wide AI agent deployment, from administration to design.
Prerequisites
This course requires the following background:
- Practical experience: 6 to 12 months of hands-on Copilot Studio experience is recommended.
- Technical skills: knowledge of Power Fx expressions and Microsoft Dataverse tables and relationships.
- Generative AI knowledge: an intermediate understanding of retrieval-augmented generation (RAG), prompt engineering, Model Context Protocol (MCP) and Agent2Agent (A2A).
- Additional skills: familiarity with REST APIs and Power Platform environments.
Teaching and assessment methods
- Initial skills assessment
- Training materials provided to participants
- Continuous assessment throughout the course
- End-of-course feedback questionnaire
- Combination of theory and practical application
- Attendance records
- Post-course follow-up evaluation
- Practical exercises
Course highlights
- A complete pathway: progress from Copilot administration fundamentals to advanced agent design in 4 days without taking two separate courses.
- Certified expert trainer: learn from a Microsoft-certified trainer with production experience in Copilot Studio.
- Intensive practical approach: a lab in every module provides hands-on practice with the Microsoft 365 admin centre and complete agents using RAG, MCP and multi-agent orchestration.
- AB-620 certification examination included: the Microsoft Certified: AI Agent Builder Associate examination is included. Microsoft Learn may also be consulted during the examination.
Dates and sessions
Choose the date and delivery format that suit you.
No upcoming sessions are currently available.
Microsoft®, Microsoft 365®, Microsoft Copilot® and Microsoft Copilot Studio® are registered trademarks or trademarks of Microsoft Corporation in the United States and other countries.