Introduction
On September 17 2025 I gave a lecture about Model Context Protocol for the New York Open Statistical Programming meetup. The primary directive I was given from organizer Jared Lander was to “teach us something new.”
My internal goals for the talk were:
- Define the main components of an MCP architecture
- MCP’s architectural value is that it standardizes what was once a tedious integration problem into a single way of doing things.
- Dive into example implementations to differentiate agentic AI vs. automation workflows.
As MCP was introduced in November 2024 this subject presented a fascinating opportunity to deepen my own understanding and a challenge of standards that are still in flight. Below please find the, slides, video and GitHub repository.