Large language models are powerful, but without access to your data and tools, they’re blind. The Model Context Protocol (MCP) is an open standard that bridges that gap, allowing LLMs to securely interact with your APIs, databases, and internal systems.
In this full-day hands-on workshop, you’ll learn how to turn your existing services into MCP servers that can power AI-enhanced tools, chat interfaces, and developer assistants. We’ll go from concept to code, exploring the building blocks of MCP servers, the different MCP types (Tools, Resources, and Prompts), and how to host and connect them to clients like VS Code or other custom AI copilots.
The second part of the day focuses on the consumer side: if you’re building your own AI-powered chat, agent, or assistant, you’ll learn how to connect to and consume MCP servers to give your LLM access to structured, trusted data and capabilities. You’ll see how MCP can unify access to multiple systems and make your assistants context-aware, without hardcoded logic or credentials.
What you’ll learn:
• How to build your first MCP server from an existing API
• Implementing and describing Tools, Resources, and Prompts
• Structuring complex types and metadata for client consumption
• Hosting and security options
• How to consume MCP servers from your own AI chat or assistant
• Integrating MCP with AI tools like VS Code, ChatGPT, and Azure AI

We’ll work primarily in .NET 10 and C#
Demos and exercises will run locally using VS Code, and some resources run in Azure.