The Universal Plug for AI? How MCP Servers Are Changing the AI Landscape

AI Just Got Its Own Universal Adapter

Remember when every phone manufacturer used a different charging cable? Those days are mostly behind us, thanks to USB-C — one connector that works with almost everything. MCP servers (Model Context Protocol Servers) are essentially the USB-C port of the AI world.

In November 2024, AI company Anthropic introduced MCP — a standardized set of rules that governs how AI models connect to external data, tools, and applications. For example, if you ask an AI to "organize my meetings for today," MCP servers allow it to directly access your Google Calendar and pull the information it needs. Previously, developers had to build custom integrations for every single service. Now, once the standard is in place, it works everywhere.

How Do MCP Servers Actually Work?

Here's a simple way to think about it: the AI (the client) says, "I need this data," and the MCP server acts as the middleman, responding, "Got it — here you go."

MCP servers provide three main things:
- Tools: Actions the AI can perform (e.g., sending an email, running a search)
- Resources: Data the AI can read (e.g., company documents, databases)
- Prompts: Pre-built question templates for common use cases

There are currently over 10,000 MCP servers running worldwide, with monthly downloads exceeding 97 million. ChatGPT, Claude, Microsoft Copilot, and Google Gemini all officially support MCP.

A New Zealand Company Uses MCP to Reimagine Data Management

New Zealand-based data management firm RecordPoint recently launched its own MCP server. It connects sensitive enterprise data to AI tools while maintaining security and regulatory compliance.

Joseph Pearce, RecordPoint's Head of Product, explained: "When companies bring in AI agents, their biggest concerns are security and compliance — and MCP servers address those issues in a standardized way." With 67% of New Zealand's large enterprises already using AI in their operations (as of 2024), the spread of standards like MCP is expected to broaden the scope of AI adoption even further.

South Korean Companies Are Experimenting with MCP Too — The Baemin Case

In South Korea, Woowa Brothers — the company behind Baemin (Korea's leading food delivery app) — implemented an MCP server at an internal developer hackathon in April 2025. A team of six developers and two product managers built a feature in a single day: an AI that searches for products on Baemin's grocery service and automatically compares prices across different sellers.

Here's how it works: a user types something like "I want to buy eggs, milk, and onions," and the AI uses the MCP server to access product data and instantly recommend the seller with the lowest price. This is a prime example of how MCP servers are becoming the connective tissue that lets businesses quickly layer AI capabilities onto existing services.

Key Takeaways

MCP servers enable AI to connect with a wide range of apps, data, and tools through a standardized protocol.
Introduced by Anthropic in November 2024, MCP has since been adopted by all major AI players including OpenAI, Google, and Microsoft.
Over 10,000 MCP servers are currently in operation worldwide, with more than 97 million downloads per month.
New Zealand's RecordPoint launched an MCP server that securely connects enterprise data to AI tools.
South Korea's Woowa Brothers built an AI-powered price comparison feature for their grocery service in just one day using MCP at a hackathon.

Wrapping Up

MCP servers are becoming a key enabler in the shift from AI as a "chat companion" to AI as a genuine business tool. Whether you're in New Zealand or South Korea, if you want to use AI smarter and more effectively, MCP is a term worth keeping on your radar.