Updated July 2026

Building Agents with Microsoft 365 Copilot

Class Duration

7 hours of live training delivered over 1-2 days to accommodate your scheduling needs.

Student Prerequisites

Comfort with everyday office software: documents, spreadsheets, email, and a web browser
Familiarity with Microsoft 365 apps such as Teams, SharePoint, and Outlook
No technical background or programming experience required

Target Audience

Business users and team leads who want to build their own AI assistants inside Microsoft 365: operations, HR, finance, marketing, sales, support, and management. Ideal for organizations that have adopted Microsoft 365 Copilot and want staff to create purpose-built agents grounded in their own documents and processes. No technical background is assumed, and nothing in the course requires one.

Description

This practical, hands-on course teaches business users how to build their own agents in Microsoft 365 Copilot, no programming required. Participants learn what a Copilot agent is, how it differs from ordinary Copilot chat, and how to create one using the built-in agent builder. The course covers writing clear instructions, grounding an agent in trusted knowledge from SharePoint sites and uploaded files, adding actions so an agent can do more than answer questions, and testing an agent until it behaves reliably. Sharing and governance receive close attention: how to publish an agent to a team or the whole organization, how Microsoft 365 permissions and data boundaries protect information, and how to work within your organization's policies. Every topic is practiced against realistic workplace scenarios so participants leave able to build, refine, and share useful agents the same week.

Learning Outcomes

Explain what a Microsoft 365 Copilot agent is and how it differs from standard Copilot chat.
Create a custom agent with the no-code agent builder from a plain-language description.
Write clear instructions that shape an agent's role, tone, scope, and behavior.
Ground an agent in trusted knowledge from SharePoint, files, and approved sources.
Add actions and connectors so an agent can retrieve information and complete tasks.
Test and refine an agent, checking its answers against its knowledge sources.
Publish and share an agent with a team or the wider organization.
Apply Microsoft 365 permissions, data boundaries, and organizational policy to keep agents safe.
Decide which everyday tasks are good candidates for a purpose-built agent.

Training Materials

Comprehensive courseware is distributed online at the start of class. All students receive a downloadable MP4 recording of the training.

Software Requirements

Students need a Microsoft 365 account with access to Microsoft 365 Copilot and the Copilot agent builder, plus a web browser. No installation required.

Training Topics

Understanding Copilot Agents

What an agent is and how it extends Microsoft 365 Copilot
Agents compared to ordinary Copilot chat
Where agents appear: Teams, Copilot Chat, and Microsoft 365 apps
Everyday scenarios where a purpose-built agent helps
The lifecycle of an agent from idea to shared tool

Creating Your First Agent

Starting from a plain-language description in the agent builder
Naming an agent and writing its description
Setting the agent's role, scope, and personality
Adding conversation starters that guide users
Previewing the agent as you build

Writing Effective Instructions

The anatomy of clear agent instructions
Defining what the agent should and should not do
Setting tone, audience, and response format
Handling questions outside the agent's scope
Refining instructions based on how the agent responds

Grounding an Agent in Knowledge

Connecting an agent to SharePoint sites and document libraries
Uploading files as a knowledge source
Pointing an agent at approved web content
Keeping knowledge current and trustworthy
How grounding reduces made-up answers

Adding Actions and Connectors

What actions let an agent do beyond answering questions
Connecting to Microsoft 365 data and approved services
Choosing prebuilt connectors for common tasks
Understanding what an action can access on your behalf
Reviewing an action's results before relying on them

Testing and Refining

Trying an agent against realistic questions
Spotting weak instructions and knowledge gaps
Checking answers against the agent's sources
Iterating on instructions, knowledge, and actions
Getting an agent ready to share

Publishing an agent to yourself, a team, or the organization
How Microsoft 365 permissions and data boundaries apply
The role of administrators and approval in publishing
Protecting sensitive and confidential information
Working within your organization's AI and data policies

Building Agents That Last

Spotting agent opportunities in your own workday
Keeping instructions and knowledge maintained over time
Gathering feedback from the people who use your agent
Sharing effective agent practices across your team
Staying current as Copilot capabilities change