The illustrations were created in cooperation between humans and artificial intelligence. They show a future in which technology is omnipresent, but people remain at the center.
AI-generated illustration
Microsoft Copilot Studio: Build Your Own AI Agents Without Coding
A Practical Approach: From Use Case to Concrete Pilot Agent and Beyond
Online
2 days
German
€ 1.390,-
plus VAT.
€ 1.654,10
incl. VAT.
Booking number
43276
Venue
Online
.svg)
5 dates
€ 1.390,-
plus VAT.
€ 1.654,10
incl. VAT.
Booking number
43276
Venue
Online
5 dates
plus VAT.
incl. VAT.
Booking number
Venue
0 Dates
Become a certified
Machine Learning Engineer
Machine Learning Engineer
This course is part of the certified Master Class "Machine Learning Engineer". If you book the entire Master Class, you save over 15 percent compared to booking this individual module.
To the Master Class
In-house training
In-house training just for your employees - exclusive and effective.
Inquiries
In cooperation with
In cooperation with
AI agents are considered the next big step after chatbots—but there’s a lot of work involved between “I’ve opened Copilot Studio before” and an agent that’s actually operational and secure. This two-day “ training ” provides hands-on guidance, taking you from selecting a suitable use case to building a demo-ready agent in Microsoft Copilot Studio.
Over the course of both days, you’ll build your own end-to-end agent: You’ll connect knowledge sources via retrieval-augmented generation, design controlled dialogs, integrate a specific action, and learn about the fundamental requirements for permissions, security, governance, and operations. More advanced topics such as complex APIs, MCP, CI/CD, or Git integration are intentionally introduced only in context and not covered in depth—the focus remains consistently on a practical introduction.
By the end, you’ll be able to realistically assess potential use cases for agents, build your first agent yourself, and outline the next steps for a pilot project within your own company.
Contents
1. Fundamentals and Classification of Copilot Studio Agents
- LLM Basics: A Concise Guide.
- A Comparison of M365 Copilot, Agent Builder, and Copilot Studio Agents.
- Entry Points and Types of Agent Triggers.
- Knowledge sources, topics, and actions as core building blocks.
- Identify and define suitable use cases.
2. Create Your First Agent
- Navigating the Copilot Studio Interface.
- Create your first agent.
- Define the role, tone, boundaries, and instructions.
- Test and customize the agent in the editor.
3. Knowledge and Retrieval-Augmented Generation (RAG)
- Basic Principles of RAG and the Role of Knowledge Sources.
- Make effective use of typical knowledge sources such as SharePoint, files, and websites.
- Grounding, permissions, and source quality as factors for success.
- Deliberately limit hallucinations.
4. Topics, Guided Dialogue, and Guardrails
- Distinguish clearly between topics and generative responses.
- Triggers, questions, variables, and conditions as control elements.
- Create a topic for a clearly defined process.
- Define test questions, success criteria, and guardrails for data access and behavior.
5. Integrate Actions into the agents
- Suitable use cases for actions—and how they differ from traditional flows.
- Configure income and expenses; schedule user confirmations.
- Integrate a simple action into your own agent.
6. Using Agent Flows
- Typical use cases for agent flows.
- Configure inputs, outputs, and parameters correctly.
- Structure multi-step processes and make them easy to follow.
- An overview of the capabilities of API, MCP, and the GitHub Copilot agent.
7. Publish, test, and further develop agents
- Classify permissions, user context, and appropriate channels.
- Prepare for publication and approve your own agent.
- Monitor usage and response quality with analytics.
- Run the tests again and evaluate the improvements in a way that is easy to understand.
8. Enterprise Readiness and Your Own Pilot Agent
- Provide a concise overview of governance, data protection, responsibilities, and the support model.
- Environments and solutions as the foundation for secure operations.
- An Overview of Licensing, Capacity, and Billing Models.
- Create and pitch your own 90-day action plan for the pilot.
Your benefit
- You'll build your own, fully functional AI agent in Microsoft Copilot Studio—from the initial idea to release.
- You securely integrate knowledge sources such as SharePoint or documents using Retrieval-Augmented Generation and effectively limit hallucinations.
- You design controlled dialogs around specific topics and know when generative responses are more appropriate.
- You integrate Actions and Agent Flows to connect your agents to real-world processes.
- You are familiar with the governance, security, and operational requirements for production deployment.
- You use Analytics to evaluate your agent's performance and the quality of their responses, and then make targeted improvements.
- By the end of the course, you'll have a concrete 90-day action plan for launching your own pilot program within your company.
trainer
Martin Rupprecht
Methods
The course follows a "guided build" format: Over the course of both days, you'll continue to develop a single, end-to-end agent, rather than watching many independent demos.
- Brief theoretical insights with immediate practical application
- Transfer Questions After Each Hands-On Session: Data, Risk, Owner, and Success Criteria
- Live demos for topics that do not require all participants to configure them themselves
- Final Pitch for Your Own Pilot Use Case
Prerequisites: A basic understanding of Microsoft 365 is sufficient—no programming knowledge is required. A modern browser, a stable internet connection, and, ideally, a second screen for the exercises are recommended. A demo environment with licensed accounts will be provided.
Final examination
Recommended for
- Subject matter experts and power users who want to set up agents for knowledge access or departmental processes
- IT and digital transformation leaders who take a holistic approach to implementation, security, and scalability
- Process and automation managers who want to connect agents to Power Automate
- AI officers and innovation managers who want to evaluate pilot use cases in a structured manner
.svg)