Building MCP Servers Is Simpler Than Everyone Says
My AI courses: https://edwarddonner.com/curriculum
00:00 Can You Build an MCP Server Easily?
00:20 What Can an MCP Server Actually Do?
01:13 How Do You Write an MCP Tool?
02:36 How Do You Turn Python Code into an MCP Server?
03:42 What's the Cheapest Way to Deploy an MCP Server?
04:25 How Do You Deploy an MCP Server with Fly.io?
05:26 How Do You Connect an MCP Server to ChatGPT?
06:55 Does ChatGPT Actually Use Your MCP Server?
07:43 How Do You Monitor a Live MCP Server?
08:02 What's Next After Building Your MCP Server?
___
Repo is here; the README has the instructions for the 3 steps:
https://github.com/ed-donner/faq
And the MCP server that I make is running here:
https://ed-donner-faq.fly.dev/mcp
I deployed a Remote MCP Server and plugged it into ChatGPT in under 10 minutes. The hosting cost is a few cents a month. This video shows how.
MCP makes it easy to use tools created by someone else. It also makes it easy for others to use your tools. But first you need to make and deploy your own MCP Server. That’s the part that sounds hard. But that’s a misconception. It’s easy!
I create and deploy an MCP Server over Streamable HTTP on fly.io.
I then have a pleasing conversation with ChatGPT in which it automatically uses my tools several times via my remote MCP server.
There is still a hard part, though: coming up with an idea for tools worth sharing.
That’s the challenge I give you on the video. Build something useful, post your MCP URL, and I’ll be one of the first to try it. Because that part is easy.
___
MCP Technologies
Model Context Protocol (MCP)
Remote MCP Server
Streamable HTTP Transport
MCP Tools
Development Tools & Frameworks
FastMCP
Docker
Dockerfile
uv
Fly.io
Supabase
Cloud & Deployment
Fly.io
Docker Containers
Cloud Deployment
AI Development Concepts
Remote MCP Deployment
Tool Calling
Agent Development
AI Tool Sharing
Function Calling
AI Models & AI Applications
ChatGPT
Claude Desktop
Large Language Models (LLMs)