Open Workshop: Hands-On MCP
Overview
This workshop will give you a flavour of our new AI engineering programme (see below) and covers Anthropic's Model Context Protocol—a framework for connecting AI agents to external systems.
Background
MCP (Model Context Protocol) was introduced by Anthropic in November 2024 as an open standard that solves the N×M integration problem, eliminating the need to build custom connectors for each AI system-data source pairing. Instead of vendor-specific solutions like OpenAI's function-calling API or ChatGPT plugins, MCP provides a universal interface: developers build one MCP server that works across AI systems, while AI applications connect to any MCP-compatible data source without custom code.
MCP enables AI applications to interact directly with development and production environments. Developers connect AI coding assistants to IDEs, GitHub repositories, Slack, Google Drive, Postgres databases, and Stripe. Custom MCP servers extend this to proprietary systems, including inventory management, CRMs, internal dashboards, analytics platforms, and scheduling tools.
Using MCP, you can build AI applications that don't just talk about things in context; they can read codebases, query databases, check messages, send emails, and execute workflows.
What we will cover
Opening talk: MCP use cases and architecture
Build a starter MCP server
Build your own useful integration
Share & discuss
What to bring
Your laptop
Install Claude Desktop
Optional: Claude Pro subscription OR Claude API credits
Hot-desking
You are welcome to come early and hot-desk in the main classroom before the workshop.
AI engineering Q&A from 5pm
For those who are interested, we will be answering questions about our new AI engineering programme at the end of the workshop.
Other dates