What is MCP (Model Context Protocol) and why it changes how I use AI

I've been using MCP (Model Context Protocol) in my day-to-day work with AI for a while, and it has changed how I work quite a bit. Here it is without the hype.

The problem it solves

A language model, however good, only knows what you put in the prompt. It can't read your files, query your database or call your APIs… unless you give it tools. And everyone solved this their own way, with no standard.

What MCP is

MCP is an open protocol that standardises how a model connects to tools and data sources. Instead of bespoke integrations for everything, you expose an "MCP server" offering resources and tools, and any compatible client (an assistant, an IDE…) can use them.

Think of it as "USB-C for AI": a common connector between the model and the outside world.

How I use it

• Give the assistant controlled access to a project: read code, search, run commands.
• Connect data sources (docs, tickets, a database) so it answers with real context.
• Build agents that actually do things, not just write text.

Where judgement comes in

Giving a model tools is powerful and also delicate: permissions, what it can and can't touch, and validating what it does. As with everything in AI, the tool speeds you up, but the control stays yours.

By the way, I built an MCP server for this very site: you can connect and read it from your AI. I explain how on this page.

Want to build agents or bring AI into your product? Reach out from the home page.