Microsoft Copilot Studio for Beginners 2026 – Complete Guide

Creating AI Agents in Microsoft Copilot Studio

Introduction to Copilot Studio

• The tutorial begins with an overview of how to access Microsoft Copilot Studio at CopilotStudio.microsoft.com.

• Users are guided to create an AI agent by selecting "agent" from the left menu and clicking "create plank agent" in the upper right corner.

Setting Up the IT Support Agent

• The focus is on building an IT support agent capable of searching a knowledge base, utilizing both generative and rule-based actions, such as updating IT service boards.

• The agent is named "IT support agent," with a description emphasizing its role in providing self-service IT support for common issues.

Customizing the Agent

• Users are encouraged to personalize their agents by adding icons for easier identification among multiple agents within Copilot Studio.

• Instructions are provided on how to upload lesson files, including icons, which can be obtained via a link shared in the video description.

Testing Initial Functionality

• A demonstration shows how to test the agent's responses using prompts like “my VPN stopped working after I changed my password,” highlighting that it currently lacks knowledge.

• The initial web search results indicate that the AI pulls information from various sources, which may not be ideal for specific IT support scenarios.

Improving Knowledge Sources

• To enhance performance, users are advised to select appropriate models; experimental models should be avoided due to instability.

• Users learn how to disable web search capabilities and add knowledge directly from OneDrive or other sources instead of uploading files manually.

Adding Knowledge Base Content

• A step-by-step process is shown for copying relevant documents into OneDrive and creating a new folder specifically for IT support knowledge.

• After confirming selection of the document as a primary knowledge source, users can add this content directly into their agent’s database.

Final Testing and Validation

• Once knowledge is added, users can retest their agents. They observe improved responses based on uploaded documents when prompted about fixing VPN issues after password changes.

• Further testing includes asking about resetting passwords for different services like Salesforce, demonstrating enhanced functionality through updated knowledge integration.

How to Configure an AI Agent for IT Support

Adjusting General Knowledge Settings

• The speaker discusses the limitations of the document regarding general knowledge responses generated by the AI model, specifically GBT 5.

• Instructions are provided on how to navigate settings to disable the use of general knowledge in favor of specific IT service desk information.

• Emphasis is placed on giving explicit instructions to the AI agent, defining its role as an internal IT service desk assistant and limiting its responses strictly to relevant topics.

Testing AI Responses

• The speaker tests the AI's response by asking about resetting a Salesforce password, expecting it not to provide guidance outside its defined knowledge base.

• The AI responds with general password guidance instead of specific Salesforce instructions, indicating that further refinement of instructions is needed.

Introducing Topics for Structured Scenarios

• The concept of "topics" is introduced as a way to guide the agent through structured scenarios, enhancing its ability to respond appropriately based on user input.

• A distinction is made between custom topics (created by users) and system topics (predefined), with a recommendation against disabling system topics.

Customizing Conversation Starters

• The speaker demonstrates how to customize conversation starters within system topics, including adding icons and modifying greetings for better user engagement.

• It’s noted that changes can be saved only after clicking away from certain fields, highlighting a minor usability detail in the interface.

Capturing User Issues for Support Workflows

• A new custom topic is created aimed at capturing user issues immediately upon their description, which will be stored for later use in support workflows.

• Detailed steps are outlined for setting up this topic, including naming conventions and triggers that invoke it based on user interactions.

• Variables are introduced as containers for values; here, a global variable named "user issue text" is created to store user messages across different topics.

This structure provides clarity on configuring an AI agent tailored specifically for IT support tasks while ensuring effective communication with users.

Creating an IT Support Request Flow

Overview of the Process

• The discussion begins with handling relations and creation of IT support requests when knowledge is insufficient, allowing the agent to choose how to proceed.

• A question is posed to the user: "Would you like me to create an IT support request for you?" This initiates a multiple-choice interaction.

• Two options are created for user response: "Yes" or "No," leading to different branches in the flow. This choice variable is labeled as v1.

Confirmation and Messaging

• If the user opts not to create a ticket, a message will be sent stating that their IT support request has been created and that the service desk will follow up shortly.

• Additional messaging is included for users who decline ticket creation, offering further assistance if needed.

Best Practices in Naming Variables

• It’s recommended to rename actions clearly for better understanding, while retaining original names during learning phases for clarity on functionality.

• The importance of consistent naming conventions (e.g., Pascal case notation) is emphasized for ease of navigation within the system.

Step-by-Step Instructions

• The process involves six steps where each step must be completed before moving on. Initial steps include capturing user issues and determining if they can be resolved using internal knowledge.

• If internal knowledge cannot resolve the issue, it prompts confirmation from the user about creating an IT support request.

Implementation of Instructions

• Instructions are edited and inserted into the system after copying them from lesson files. Placeholders are replaced with actual topics such as "capture the issue."

• After saving instructions, testing reveals that AI agents can effectively follow these steps when users describe problems like resetting passwords.

Tools Integration

• Discussion shifts towards integrating tools within the AI agent's capabilities, including custom connectors and workflows similar to Microsoft Power Automate.

• An agent flow tool is introduced which allows independent action by AI agents based on user requests, specifically focusing on ticket creation processes.

How to Use Microsoft Planner for Issue Reporting

Setting Up Microsoft Planner

• The speaker introduces the use of Microsoft Planner, which is available for free with a Microsoft 365 subscription.

• A new team named "issue reporting demo" is created along with a channel called "support issues."

• The purpose of creating a team is to enhance functionality and collaboration within the issue reporting process.

Creating an Issue Board

• An issue board titled "IT support issue tracking" is established, featuring buckets for organizing tasks.

• The initial bucket labeled "to-do" allows users to input new issues, which can then be moved through different statuses like "in progress."

Adding Inputs and Tasks

• Users can add inputs such as an "issue description," which will be utilized in task creation.

• When creating a task in Planner, users must provide details like the title and assign it to specific individuals.

Working with Dates and Time Zones

• The speaker explains how to set start dates using UTC time, emphasizing the need for time zone conversion.

• Instructions are provided on adding due dates by incorporating expressions that calculate future dates based on current UTC time.

Updating Task Details

• To update task descriptions, another action must be added where the task ID from earlier steps is referenced.

• After setting up all necessary actions, users are encouraged to save their draft and review analytics related to tool usage efficiency.

Testing the IT Support Tool

Manual Testing of the Tool

• The testing process begins with a manual run of the tool, requiring initial sign-in and an issue description input.

• An example issue is provided: "My SAP doesn't work at all. Please help. Urgent." This serves to verify that the tool functions correctly.

• After running the flow, confirmation appears indicating successful execution, demonstrating that the tool can handle user requests effectively.

Creating IT Support Requests

• The system generates an IT support request based on user input, including details like title, start date, due date, and notes.

• There’s a need for these requests to be categorized under specific topics for better organization within the system.

Summarizing User Issues

• Users may submit lengthy problem descriptions; thus, summarization features are introduced to condense information efficiently.

• A new prompt action is created in AI Builder to summarize issues into two clear sentences suitable for IT support tickets.

Implementing Variables and Testing Prompts

• Input variables are established (e.g., "user issue text") to facilitate dynamic content handling during testing phases.

• Sample data is utilized for testing purposes; this helps ensure that prompts yield concise summaries as intended.

Finalizing and Saving Configurations

• The summary output variable ("issue summary") is prepared for integration with tools needed to create IT support tickets.

• Upon saving configurations, users can test their setup by submitting long messages which trigger automated responses from the system.

IT Support Agent Creation and Publishing Process

Debugging Tool Explanation

• The debugging tool is only available in the test window and not when publishing the agent. Users may need to refresh the page to see the IT support request.

Agent Functionality Overview

• The agent can create items and summarize user content, enhancing user interaction by allowing for new test sessions.

• It captures issues effectively by searching an internal knowledge base for solutions, although it may sometimes fail to find relevant information.

User Interaction with IT Support Request

• The agent prompts users for confirmation before creating an IT support request, ensuring clarity in communication.

• Messages are distilled from user input, which helps in generating appropriate responses or actions based on captured issues.

Publishing the Agent

• To publish the agent, users must click 'publish' twice and wait a few seconds for completion.

• Multiple channels are available for publishing, including Teams, SharePoint, WhatsApp, Facebook, Salesforce, as well as web or native apps.

Customizing Channel Settings

• Users can customize settings after adding channels; this includes changing icons and descriptions to better represent the service provided.

• Background colors can be adjusted for aesthetics; a white background is preferred over green. Descriptions should clearly state the purpose of the agent.

Finalizing Changes Before Publication

• Additional configurations include setting developer names and saving changes before final publication.

• A comprehensive Copilot Studio course is mentioned as a resource for further learning about these processes.

Adding Agent to Teams

• After confirming access permissions (e.g., OneDrive), users can add their support agent to Teams with necessary descriptions visible during setup.

Testing Functionality Post-Publication

• Once added to Teams, users can interact with the support agent directly; it confirms issue capture and offers ticket creation options seamlessly within Teams.

Sharing Options Within Organization

• Users have options to share agents either with specific teammates or across their entire organization; admin approval may be required for broader sharing.

Microsoft 365 Admin: Publishing IT Support Agent

Navigating the Microsoft 365 Admin Portal

• The speaker discusses sending a request to the IT admin for approval, emphasizing the need for improvements in the Microsoft 365 admin interface.

• The process of accessing the admin portal is outlined, where users can view pending requests for IT support agents and initiate publishing actions.

• Options are provided to restrict agent installation to specific user groups or allow all users access, showcasing flexibility in user management.

• After publishing, every user can find and interact with the IT support agent within their Microsoft 365 environment.

Interacting with the IT Support Agent

• Users can engage with the IT support agent by initiating conversations, such as reporting issues like "work day failing to load my pay slip."

• The agent prompts users to create an IT support request if their issue isn't found in existing knowledge bases, demonstrating proactive assistance.