Sorry for the clickbait. But if you’re here, it’s because someone said “MCP” in a meeting, in a LinkedIn post, or in a tweet, and you nodded along like you knew what they were talking about. Don’t worry — I’ll explain.
First, the acronym
M.C.P. = Model Context Protocol
Sounds technical. It is. But the idea is simple. Let’s break it down word by word:
Model — A word that’s already part of everyday conversation. ChatGPT, Claude, Gemini… these are models. Programs trained to understand and generate text, code, images. The “brain” of the AI you use every day.
Context — You already get this intuitively. When you talk to an AI, the more context you give it, the better it responds. It’s not the same to say “make me an ad” versus “make me an Instagram ad for a premium Colombian coffee brand targeting millennials in Buenos Aires.” Context is the information the model needs to do its job well.
Protocol — The standardized, secure, and effective way to pass that context to the model. Think of it as a universal plug. Before, every tool connected to AI in a different, improvised way. Anthropic (the creators of Claude) said: “let’s make a standard” — and MCP was born.
OK, but what’s it for?
Imagine you have Claude (or any AI) and you want it to help with your work. Today you can already ask it to write an email, summarize a document, generate code. But there’s a limit: the AI only knows what you tell it in that moment.
What if you want the AI to look at your Facebook Ads data and tell you which campaign is performing best? Or read your Notion and put together a weekly summary? Or check your Shopify and alert you if sales dropped?
That’s where MCP comes in. It’s the bridge that connects AI to your tools. It gives AI direct access to the information it needs, without you having to copy and paste like a monkey.
A concrete example
Without MCP:
- Open Facebook Ads Manager
- Export data to Excel
- Open ChatGPT
- Paste the data
- Ask it to analyze
- Get a response based on data that’s already 20 minutes old
With MCP:
- Tell Claude: “How are my Facebook campaigns doing this week?”
- Claude connects directly to your Facebook Ads account
- You get a response with real-time data
The difference isn’t just speed. It’s quality. The AI has access to real, up-to-date data, without the filter of your copy-paste.
Why now?
Anthropic launched the MCP standard in November 2024 and major companies like Slack, Notion, GitHub, and many more quickly jumped on board. In just over a year, the ecosystem went from zero to over 8,600 MCPs listed in public directories. This isn’t experimental technology — it’s infrastructure already running in production.
What happened is simple: AI stopped being a closed box that only answers questions. Now it can do things. Connect to services, read data, execute actions. MCP is what makes that happen in an orderly and secure way.
Who should care?
If you work with data, you should care. If you do digital marketing, you should care. If you run an e-commerce business, you should care. If you make decisions based on metrics, you should care.
You don’t need to be a programmer. You don’t need to understand the protocol under the hood. You need to know it exists and that the tools you use are going to start having it (or already do).
The analogy I like most
Remember when phones didn’t have apps? You had the phone and that was it. Then the App Store came along and suddenly your phone could do everything: order food, take professional photos, manage your bank.
MCP is something similar for AI. It’s what allows AI to “install” connections to the real world. Each MCP is like an app that gives AI a new capability: read your marketing data, connect to your CRM, analyze your inventory.
What does Detrics have to do with it?
At Detrics we built an MCP that connects AI to over 50 marketing and e-commerce platforms: TiendaNube, VTEX, Facebook Ads, Google Ads, TikTok, Shopify, Google Analytics, and more. Basically, we give AI the context of your business data so you can ask whatever you need without leaving where you’re working.
But that’s a topic for another article →.
If this is one of the first articles you’re reading from our blog: I’m Tom, co-founder and CTO of Detrics. I write about AI, data, and the tools we’re building to make working with data less painful.
If you want to understand how MCPs are changing reporting, read: Why MCPs Are the Future of Reporting →