This five-day course introduces participants to designing, building, integrating, and publishing AI agents using Microsoft Copilot Studio. Participants develop agents that answer questions using organizational knowledge, guide users through structured conversations, connect to external services, and automate business processes.

The course progresses from agent fundamentals and conversation design to REST API integration, Model Context Protocol (MCP), and automated workflows using agent flows and Power Automate. Hands-on exercises culminate in a capstone project that combines these capabilities into a practical business solution.

 

Duration 5 Days – 35 hrs.

 

Objectives

• Explain Microsoft Copilot Studio capabilities and identify suitable business use cases.
• Create AI agents with clear instructions, defined scope, and appropriate knowledge sources.
• Build conversational topics using entities, variables, conditions, and Power Fx expressions.
• Configure generative orchestration and describe tools so agents can select appropriate actions.
• Integrate REST APIs using supported API tools or custom connectors.
• Connect agents to an existing MCP server and use its exposed tools.
• Build automated workflows using agent flows and Power Automate.
• Apply authentication, access controls, data policies, and responsible AI practices.
• Troubleshoot agent responses, tool calls, and workflow execution.
• Publish and maintain an integrated agent solution.

  

Target Audience 

• Business analysts and business process specialists.
• Power Platform makers and citizen developers.
• Automation specialists and solution consultants.
• Developers integrating business applications and external services.
• IT professionals supporting Microsoft 365 and Power Platform solutions.
• Technical team members responsible for enterprise AI agent adoption.

 

Prerequisites 

• Basic familiarity with Microsoft 365 applications and web-based business systems.
• Understanding of common business processes and workflow concepts.
• Basic knowledge of data tables, variables, and conditional logic.
• Introductory familiarity with HTTP, REST APIs, and JSON; sample specifications and payloads will be provided.
• A work or school account with access to Microsoft Copilot Studio and a suitable Power Platform training environment.
• Appropriate permissions and licensing for the planned connectors, workflows, knowledge sources, and publishing channels.
• Access to prepared sample documents, a REST API with an OpenAPI specification, and a compatible remote MCP server with required credentials.
• A computer with a supported browser and reliable internet access.

                                      

  

Course Outline 

Day 1: Microsoft Copilot Studio Foundations

Module 1: Platform Overview and Agent Planning

• Microsoft Copilot Studio within the Microsoft ecosystem.
• AI agents, conversational experiences, and business automation.
• Core components: instructions, knowledge, topics, tools, and channels.
• Generative orchestration and explicitly designed conversation paths.
• Defining business requirements, agent boundaries, and expected outcomes.
• Environments, access permissions, licensing, and capacity concepts.

 Module 2: Creating an Agent and Adding Knowledge

• Navigating the authoring environment.
• Creating an agent and writing effective instructions.
• Adding supported knowledge sources, including documents, websites, and SharePoint.
• Grounding responses and handling unavailable information.
• Reviewing responses and refining agent behavior.
• Hands-on exercise: Build an internal policy assistant using prepared organizational documents.

  

Day 2: Conversation Design and Tool Orchestration

Module 3: Topics, Entities, Variables, and Conversation Logic

• Creating and managing topics.
• Trigger phrases and topic descriptions.
• Capturing user information through questions and entities.
• Topic and global variables.
• Conditions, branching, and topic redirection.
• Introductory Power Fx expressions.
• Fallback behavior and escalation paths.
• Hands-on exercise: Create a service request conversation with input validation and conditional routing.

 Module 4: Tools and Generative Orchestration

• Choosing between knowledge retrieval, topics, and tools.
• Adding prebuilt connector tools.
• Defining tool names, descriptions, inputs, and outputs.
• Configuring tool availability and invocation behavior.
• Requesting user confirmation before consequential actions.
• Troubleshooting missing inputs and incorrect tool selection.
• Hands-on exercise: Extend the agent with a connector tool that retrieves a business record.

 

Day 3: REST API Integration and Automated Workflows

Module 5: Connecting Agents to REST APIs

• REST concepts: endpoints, HTTP methods, headers, and status codes.
• JSON request and response structures.
• Understanding OpenAPI specifications.
• Adding REST API tools where available.
• Using custom connectors for supported integration scenarios.
• Authentication options and secure connection management.
• Mapping conversation variables to API inputs.
• Handling validation errors, authorization failures, and unavailable services.
• Hands-on exercise: Retrieve service request details and create a request through a prepared REST API.

 Module 6: Agent Flows and Power Automate Workflows

• Agent flows and Power Automate cloud flows.
• Agent-initiated, event-triggered, and scheduled workflow scenarios.
• Passing inputs to flows and returning results to agents.
• Conditions, loops, data operations, and connector actions.
• Approval processes and notifications.
• Handling workflow failures and preventing duplicate submissions.
• Hands-on exercise: Automate request routing, approval, record updates, and notifications.

  

Day 4: Model Context Protocol and Secure Integration

Module 7: Integrating Model Context Protocol

• MCP concepts: clients, servers, and tool discovery.
• Comparing MCP connections, REST API tools, and connectors.
• Connecting Copilot Studio to an existing remote MCP server.
• Configuring authentication and connection settings.
• Discovering and selecting exposed tools.
• Understanding tool schemas, parameters, and results.
• Controlling tool access and reviewing external tool behavior.
• Troubleshooting connection and invocation failures.
• Hands-on exercise: Connect to a prepared MCP server and retrieve asset availability through an exposed tool.

Module 8: Security, Governance, and Integration Reliability

• User authentication and Microsoft Entra ID concepts.
• User credentials and maker-provided connections.
• Least-privilege access and data protection.
• Power Platform data policies and connector restrictions.
• Prompt injection risks and untrusted external content.
• Safe handling of sensitive information and secrets.
• Confirmation steps for actions that change business records.
• Hands-on exercise: Strengthen the integrated agent’s permissions, confirmation steps, and error handling.

  

Day 5: Publishing, Operations, and Capstone Project

Module 9: Publishing and Lifecycle Management

• Publishing agents to supported channels.
• Channel configuration and access requirements.
• Development, test, and production environment concepts.
• Solutions, connection references, and environment variables.
• Monitoring conversations, tool execution, and workflow runs.
• Using analytics and feedback to improve agent behavior.
• Capacity, consumption, ownership, and ongoing maintenance.

 Module 10: Capstone Project — Integrated Employee Service Agent

• Define an employee service scenario and its business rules.
• Configure agent instructions and organizational knowledge.
• Create a structured conversation that captures and validates request details.
• Integrate a REST API to retrieve or create service requests.
• Connect an MCP tool to check asset availability or retrieve supporting operational information.
• Trigger an automated workflow for routing, approval, and notification.
• Apply authentication, access controls, and user confirmation.
• Handle missing information, failed integrations, and duplicate requests.
• Publish the solution to an approved training channel where licensing permits.
• Prepare a solution summary covering configuration, dependencies, and maintenance.

 

Capstone deliverables:

• A configured Copilot Studio agent.
• Connected REST API and MCP tools.
• An automated business workflow.
• Supporting configuration and operational documentation.