Copilot Studio for Beginners: Hands-On AI Agent Build

Name: Copilot Studio for Beginners: Hands-On AI Agent Build
Rating: 4.4 (127 reviews)

Design, test and publish your first Microsoft Copilot Studio agent with guided real-world projects, no coding required

Created byAnders Jensen

Last updated 4/2026

English

What you'll learn

Build and configure your first Copilot Studio AI agent from scratch
Connect files, websites and data sources as agent knowledge bases
Design conversation flows with Topics, variables and conditions
Trigger Power Automate actions to extend agents with real-world tasks
Publish agents to Microsoft Teams and embed them on a website
Track usage and improve performance with Copilot Studio Analytics

Course content

9 sections • 51 lectures • 3h 23m total length

Welcome0:31
Download the course materials0:02
Please find the lesson files attached to this lecture.

Get started with Copilot Studio1:25
Log in to Copilot studio at copilotstudio.microsoft.com, explore the agent builder and templates, review the quick start guide, and note the rebranding from copilot to Copilot studio.
Create and test your first agent7:28
Create and customize your first Copilot Studio agent by selecting a language, naming it, setting an icon and description, adding knowledge from Sony's public site, and testing its responses.
Publish your agent4:22

Set up the basic agent3:12
Add SharePoint knowledge4:41
Learn to create a SharePoint site, upload text and PDFs, and connect the site to Copilot Studio to add knowledge for the AI agent.
Add Dataverse knowledge5:23
Create a Dataverse table in Power Apps with columns for product ID, product name, monthly price, and yearly price. Add this insurance product prices knowledge to your Copilot Studio agent.
Add files as knowledge (+ enterprise data connections)1:43
Test the agent3:19
Synonyms2:00
Learn how to enhance an AI agent with synonyms to map cost queries to the correct data in a Dataverse MP column, enabling accurate monthly and yearly home insurance pricing.

Intro and the basic agent3:56
Build a basic travel agency agent in Copilot Studio, configure its name and description, add knowledge from a public travel site, and introduce topics to control safe-travel conversations.
System topics6:44
Explore how to manage system topics in Copilot Studio, setting triggers for conversation starts, editing start messages, switching between text and speech, and testing with quick replies and variations.
Custom topics5:45
Explore custom topics in the Copilot Studio for Beginners course to orchestrate agent conversations, configure triggers and boolean endings, and test travel destination flows like Denmark, Canada, and Australia.
Generative AI trigger2:51
Reframe the agent's trigger with generative ai settings, blending action topics and knowledge to produce safer, precise answers. Save, test, and refine prompts to identify destinations that are not dangerous.
Settings3:25
Navigate topic settings with zoom, pan, and reset; use Copilot to add steps from natural language, manage comments and variables, and verify topics with the checker.

Create a topic from blank4:24
Create a travel inspiration topic from blank, adding variables, entities, and actions, including an age question with a numeric variable named user age to tailor recommendations.
Variables5:47
Explore variables as an agent's memory by creating and naming values in string, number, and boolean types, then use the F language to build a greeting message.
Multiple choice options5:07
Capture user age in a variable, use conditional branching to greet kid if under 18 or thanks otherwise, and present a multiple choice vacation type (beach, adventure, skiing), with testing.
Conditions2:35
Master conditional logic in Copilot Studio by using variables to evaluate user vacation preference. Define conditions for beach, adventure, and skiing to branch the flow and handle outcomes.
Configure condition2:30
Configure a condition in the AI agent to trigger a beach preference with a Bora Bora image and message, then test the setup and verify the beach outcome.
Condition with an AND2:54
Explore building multiple conditions using and, or operators, activating actions by age (greater than 35), adventure conditions, and messaging within a test window for ai agent flows.
Use an API connector4:37
Understand the JSON output3:22
Use the JSON output and finish the topic6:58

Create the basic agent4:26
Create a basic agent in copilot studio that answers menu questions and takes customer orders; test it using a public menu website.
Topic intro0:44
Create Excel sheet to store orders2:29
Create topic to take orders from the customer7:10
Create a topic in Copilot Studio to take orders from customers, using generative AI in orchestration to ask for their name, log a timestamp, and request a menu item.
Integrate Power Automate7:45
Learn to integrate Power Automate with Copilot Studio to save order data to Excel, using a flow with trigger, inputs, conditions, and a discount rule.
Implement the Power Automate flow in Copilot Studio5:00
Continue the topic - Condition5:26
The next Power Automate flow9:01
Implement the next Power Automate flow8:49
Create a customer connector7:23
Implement the custom connector5:18
Create and test a custom connector in Copilot Studio to fetch the KFC menu, extract item names from JSON, and present them as a comma-separated list in a bot message.

Intro to child agents0:29
Build a reusable agent4:15
Sometimes we build one large agent and try to make it do everything. It ends up being overly complicated, vague, and hard to maintain.
Child agents fix that by splitting work into small specialist agents. So answers are sharper, updates are safer, and ownership is clear.
Each child is testable independently, and the parent routes to the right expert. Child agents come in two forms: One type is lightweight agents, and the other is normal agents, which can also be reused across multiple agents.
To demonstrate this, we will build a ‘Travel & Expenses Agent’. The parent agent has two child agents: One for travel policies and one for calculating the amount the employee must be reimbursed.
Build a parent agent3:09
Build a parent agent that orchestrates its child agents, routes travel policy questions to the travel policy agent, and routes calculations to the reimbursement agent in Copilot Studio.
Build a lightweight agent3:35
Connect the agents3:55
Reasons and tips for building child agents0:42

Publish to Teams5:35
Publish copilot studio agents to teams, copilot 365, and a web page by creating a blank agent, customizing its name, icon, and description, then publishing and adding it to channels.
Share to org and Publish to Copilot 3654:07
Publish and share Copilot Studio agents with specific teammates or the entire organization, manage availability and editor access, and navigate admin approval in Copilot 365.
Publish to website10:07
Publish a customer-facing agent to a website using Copilot Studio, configure no authentication, and embed the generated HTML or iframe chatbot into your site for live interactions.

Requirements

A Microsoft 365 tenant (trial or paid) with Copilot Studio enabled or trial activated.
Modern web browser (Edge / Chrome / Firefox) and reliable internet connection.
Basic familiarity with Microsoft 365 apps (e.g., Teams, SharePoint) is helpful but not required.
No coding or previous Power Platform experience needed, all steps are shown on screen.

Description

Microsoft Copilot Studio is Microsoft’s intuitive canvas for building AI Agents that answer questions, run business processes and sit natively inside Microsoft 365. Demand for people who can create and manage these Copilot Studio solutions is growing fast. This beginner-friendly course turns theory into practice from the very first lesson: every concept is demonstrated with a live build that you can follow click by click inside your own tenant. By the end you will have a fully published agent, real automation flows, and the confidence to repeat the process for any use case.

Why this course stands out

Hands-on from start to finish – no slide decks, every topic is taught while we build together. You watch a short demo, then pause the video and complete the same steps yourself.
Real-world projects – you will connect SharePoint files, public websites and structured data as knowledge, then trigger Power Automate to update spreadsheets, send emails and call external APIs.
End-to-end publishing – learn how to secure, test and release your AI Agents to Microsoft Teams, to a public website and to Copilot for Microsoft 365, so users can benefit immediately.
Data-driven iteration – use built-in analytics to track sessions, engagement and satisfaction, then refine your agent based on evidence instead of guesswork.

You will learn to

Create and configure a Copilot Studio agent from a blank canvas.
Ground answers in trusted knowledge sources without writing code.
Design rich conversation flows using Topics, variables and conditions.
Call external actions through Power Automate to perform real business tasks.
Publish and manage channels, privacy settings and access controls.
Read analytics dashboards to improve performance over time.

Who should enroll

Business users, citizen developers and IT professionals who want to prototype and launch AI Agents inside Microsoft 365 without programming. No prior AI background is required, only curiosity and a willingness to build along with the lessons.

Prerequisites

A Microsoft 365 account with a Copilot Studio license or free trial.
Reliable internet and a modern browser.
Optional Power Automate (free or Developer environment) to complete the automation labs.

If you learn best by doing and want to add practical Copilot Studio skills to your portfolio, join today. Together we will turn ideas into working AI Agents that save time, delight users and showcase the power of Microsoft’s latest technology.

Who this course is for:

Business users, citizen developers and IT pros who want to design, test and publish AI chat agents in Microsoft 365 - without writing code.

Copilot Studio for Beginners: Hands-On AI Agent Build

What you'll learn

Explore related topics

Course content

Intro2 lectures • 1min

Create your first Agent3 lectures • 13min

Knowledge6 lectures • 20min

Conversational Orchestration5 lectures • 23min

Advanced Conversational Orchestration9 lectures • 38min

Actions11 lectures • 1hr 4min

Child agents6 lectures • 16min

Publish3 lectures • 20min

Analytics6 lectures • 9min

Requirements

Description

Who this course is for: