// On stage
Talks
I speak at conferences, Microsoft communities and user groups across the German-speaking
tech scene — mostly in German — about agentic coding, GitHub Copilot and MCP.
01
Token Usage
EN
Where Did All My Credits Go? Understanding and Reducing Token Usage in GitHub Copilot
Since June 1, every Copilot interaction is billed by tokens, and the internet is full of tricks
that promise 60, 70, even 95 percent savings. But which of them hold up? In this talk, I explain
how tokens and AI credits actually work, show how to measure your own usage with tools you already
have, and put several popular reduction techniques to the test on real tasks from my own
repositories — with honest numbers instead of headline percentages. You'll leave knowing where
your credits go and how to check any savings claim yourself.
02
Agent Framework
DE
Designing AI-Enabled .NET Applications with Microsoft Agent Framework
Adding AI to .NET applications is fairly easy today — for example, by calling large language
models directly. The real challenge begins where AI is meant to be part of the application logic,
not just a one-off helper. This talk introduces the Microsoft Agent Framework, which helps embed
AI meaningfully beyond simple prompt calls. The focus: how to combine AI with existing .NET code,
where to draw the line between deterministic logic and AI-driven behaviour, and which architectural
considerations matter. Rather than treating AI as a replacement for established software-engineering
principles, it looks at practical patterns, trade-offs and limitations of using AI in modern .NET
applications — so you leave with a realistic sense of when the Microsoft Agent Framework is a
sensible addition to your architecture and when simpler approaches are the better choice.
03
Copilot Memory
EN
GitHub Copilot Memory: From Black Box to Memory Bank
GitHub Copilot Memory promises to give your AI a long-term memory. But what does it actually
remember, and who decides? In this talk, I explore the official Copilot Memory feature, a
cloud-hosted black box that learns automatically but offers no user control, and contrast it
with the Memory Bank pattern: a set of Git-tracked Markdown files that Copilot reads and updates
with every task. No plugins, no external tools. Just a copilot-instructions.md and a folder of
Markdown files that give your AI the context it needs to stop forgetting.
04
Pair Programming
EN
GitHub Copilot in Visual Studio: AI as a Pair Programmer
How can GitHub Copilot be used efficiently in a .NET environment? This session shows how
developers can effectively integrate GitHub Copilot into their daily workflows in Visual Studio
2026. Matthias demonstrates step by step how GitHub Copilot supports writing, refactoring, testing,
and documenting code, and how agentic coding with Agents and Instruction Files can be used
productively. Participants will receive practical best practices for using GitHub Copilot
consciously and effectively in their own .NET projects.
05
MCP Servers
EN
From Agents to Systems: Building MCP Servers for Practical AI Agents
AI agents become truly useful when they can interact with real systems rather than operate on
text alone. This session focuses on how the Model Context Protocol (MCP) can connect AI agents to
external tools and domain logic in a structured, maintainable way. It introduces the core concepts
behind MCP and explains how an MCP server is designed, implemented, and exposed to AI agents. Using
a practical example from a board game platform, the session shows how domain-specific functionality
can be made available through well-defined tools and context, enabling meaningful interactions
beyond simple prompt-based responses. The talk also covers architectural considerations such as
hosting, the boundaries between agents and systems, and common pitfalls when integrating agents
with existing applications.
06
Best Practices
DE
Effiziente Entwicklung mit GitHub Copilot in Visual Studio (Efficient Development with GitHub Copilot in Visual Studio)
How can GitHub Copilot be used really efficiently in a .NET environment? This session shows how
developers integrate GitHub Copilot into their daily workflows in Visual Studio 2026. Step by step,
Matthias demonstrates how Copilot supports writing, refactoring, testing and documenting code, and
how agentic coding with agents and instruction files can be used productively. Attendees get
hands-on best practices for using GitHub Copilot consciously and effectively in their own .NET projects.
07
MCP in .NET
DE
Microsoft Session of AI: Develop your own MCP Server
This session covers the fundamentals of the Model Context Protocol (MCP) and why it plays a central
role in the modern Microsoft AI ecosystem. We look at how an MCP server is structured, which
components it needs and how requests and responses work. Then we build our own MCP server in .NET
step by step — architecture, common implementation patterns and the most important API elements.
Finally, I show how to test the server locally and plug it into your own AI workflows.
08
AI-Driven Dev
DE
AI-Driven Development with GitHub Copilot in Visual Studio
In this talk I show how I actually use GitHub Copilot in Visual Studio in everyday .NET work —
beyond basic autocompletion. Many developers know Copilot by now, but few use it consistently and
deliberately in their daily development process. I share how Copilot helps with typical tasks, how
to collaborate with it sensibly, and how good prompts make a noticeable difference. Instead of
theoretical concepts, it's about real-world experience, common pitfalls and concrete approaches to
using Copilot more productively and purposefully.
09
Secure AI
DE
Azure Key Vault & Azure OpenAI: Secure AI Workflows in .NET
AI features are increasingly integrated into .NET applications, and every request to Azure
OpenAI can contain sensitive data. This session shows how Azure Key Vault can manage credentials
and configuration for Azure OpenAI cleanly and securely. In a live demo, a .NET application
connected to Azure OpenAI is migrated from AppSettings secrets to Azure Key Vault. Attendees
learn how managed identities, secret management and access control work together — and how to
integrate AI features into existing applications responsibly and reliably.
