Large language models are powerful, but out of the box they only know what they’ve been trained on. What if you could give an LLM new skills, access to external systems, or specialized domain knowledge, all while retaining control over the logic? That’s where Model Context Protocol (MCP) servers come in. MCP servers let you expose functionality to an LLM, enabling it to perform tasks it couldn’t do natively.

In this session, we’ll explore how to design and build a custom MCP server and integrate it with an LLM. You’ll see how MCP servers can augment AI capabilities, provide structured access to data and tools, and enable developers to safely expand the scope of what LLMs can do. We’ll also discuss best practices for building reusable skills and integrating MCP servers in various scenarios.

Attendees will leave with a clear understanding of how MCP servers can enhance LLM applications, practical strategies for building their own skills, and inspiration for adding controlled, reusable capabilities to AI-driven software.