:format(webp))
Your AI assistant is incredibly smart. It can write your emails, summarise lengthy reports, create healthy meal plans and answer your existential questions at 3am. But ask it to actually do things in the real world – like book a meeting, send an SMS or pull data from your CRM – and it hits a wall.
It's a real limitation of AI, but there's a way to break down that barrier: MCP.
What is MCP?
MCP stands for Model Context Protocol. Developed by Anthropic, it's a framework that allows AI models to connect directly to external tools, platforms and data sources.
In plain English, MCP is the bridge between your AI assistant and the rest of your software.
Without it, AI models are essentially isolated. They can think, reason and generate – but they can't actually do anything in the real world. MCP gives AI assistants a common language to talk to other software and take real action on your behalf, without needing custom code every single time.
How does MCP work?
:format(webp))
Your AI assistant lives inside its own interface: Claude, ChatGPT, whatever you use. On the other side, you have all your tools, like your CRM, calendar, messaging platform or database.
Normally, those two worlds don't talk to each other. Your AI can't see your CRM. Your CRM can't hear your AI.
An MCP server acts as a translator. It's a small piece of software that connects your AI assistant to a specific tool and gives them a common language so they can actually talk to each other.
Once the connection is made, your AI assistant can interact with that tool directly, through nothing more than a conversation. You ask. It does.
And because MCP is a shared standard, the same approach works across hundreds of different tools.
What can you use MCP for?
Once your AI assistant is connected to a tool via MCP, the possibilities are pretty broad. Anything you'd normally have to log into a platform to do manually, your AI can potentially handle through a conversation.
Some of the most common use cases:
Messaging and communications
Send SMS with your MCP, check delivery history and manage contacts. (More on this shortly.)
Calendar and scheduling
Check availability, book meetings or reschedule appointments directly in your calendar.
CRM and sales
Pull contact records, update deal stages or log notes in your CRM.
File management
Search, summarise or organise files across Google Drive or other storage tools.
Databases and data retrieval
Query databases, pull reports or surface specific records without writing a single line of SQL.
Customer support
Connect your AI to a helpdesk tool and let it draft responses, pull ticket history or flag urgent issues automatically.
Some surprisingly cool things people are doing with MCP
Once you give an AI assistant the ability to interact with the real world, people get creative. Here are some real-life examples from Reddit, that range from useful to unhinged.
Smoking a turkey
One developer built an MCP server for their smart BBQ smoker. Their AI agent monitors probe temperatures every five minutes, adjusts the fan speed and alerts them when it's time to wrap the legs. It almost worked flawlessly.
Asking your health data a question
Apple Health exports years of data points. One developer connected it to Claude via MCP so they can now just ask questions like: "Why did I sleep terribly in March?" and get actual answers, not just charts.
A wardrobe that checks your calendar
Someone built a smart wardrobe app with an MCP server that cross-references your schedule and picks appropriate outfits. Whether it always gets it right is another matter entirely.
The blog that writes and publishes itself
One developer connected research, writing and publishing tools via MCP. They give it a topic, it researches, writes and posts to their blog – in a single prompt. No editing, copy-pasting, or logging into a CMS required.
How to send SMS with an MCP Server
And then there's SMS. One of the most practical – and immediately useful – things you can do with MCP is connect your AI assistant to your messaging platform.
ClickSend's MCP Server connects to Claude, LangChain or any MCP-compatible AI tool. Then you can get your AI assistant to manage parts of your business messaging for you – everything from sending an SMS and following up leads to pulling delivery stats and working out campaign pricing.
All you need to do is ask.
Think about what that actually means in practice. You come back from a trade show with a spreadsheet full of contacts. Instead of building a campaign, crafting a message and manually hitting send, you just upload the list, tell your AI to follow up with everyone and it handles it. Or your team meeting just moved and you need to notify 50 people in a segment across three time zones. Two seconds – done.
It's not replacing your messaging platform – it's giving you a faster, more conversational way to use it.
ClickSend's MCP Server is free to set up and works with your existing account.
Who is the ClickSend MCP Server for?
You might assume that MCP is a developer thing. And setting up an MCP server for the first time does take some technical know-how. But once it's connected, the person using it doesn't need to know anything about what's happening under the hood. It's especially useful for:
Sales teams who need to follow up leads fast
Ops and support teams handling high volumes of inbound messages
Marketing coordinators who pull campaign stats and reports by hand every week
Developers who want to add messaging capability to their AI stack without extra code
Any sole trader or small business that regularly sends SMS but would rather just ask than log in
How to connect Claude to the ClickSend MCP Server
:format(webp))
Getting set up takes about five minutes – even if you’re not tech-savvy. Here’s how to do it:
1. Create a ClickSend account
If you don't have one yet, sign up free at dashboard.clicksend.com/signup
2. Download Claude
Head to claude.com/download and select your operating system.
3. Install the ClickSend MCP Server
Grab the latest version from NPM or download from GitHub.
4. Open your Claude settings
Go to Settings from the system menu bar, navigate to the Developer tab and click Edit Config. This opens a configuration file on your computer.
5. Copy and paste this into the config file
{
"mcpServers": {
"clicksend": { "command": "npx",
"args": ["-y", "@clicksend/clicksend-mcp-server@latest"],
"env": {
"CLICKSEND_USERNAME": "<your username>",
"CLICKSEND_API_KEY": "<your API key>"
}
}
}
} 6. Add your ClickSend credentials
Replace <your username> and <your API key> with your actual details. Find them at dashboard.clicksend.com/home
7. Restart Claude
You should be all connected. Look for the hammer icon – that's how you know it's working.
What's next for MCP?
MCP is still relatively new. The number of available MCP servers is growing fast – and so is the number of AI tools that support the protocol. What feels novel today will feel normal in a few years. The businesses getting familiar with it now will be well ahead of the pack when that happens.
For most people, the easiest place to start is somewhere practical and immediate, like SMS. Connect your AI assistant to ClickSend, send your first message in plain English and see how satisfying it feels when your AI actually gets something done for you in the real world.
From AI that thinks to AI that does – welcome to the next era.