Demystifying the MCP Craze

Introducing the latest entry in the AI jargon box: MCP, or Model Context Protocol. Let's cover the history of AI tool calling, why everyone is going crazy over MCPs... and is it really the "USB C of AI"?

How did we get here?

Imagine your favourite AI assistant. It can answer your questions, write emails, and even help you brainstorm ideas. But here's the thing: on its own, an AI model is like a brain that's incredibly knowledgeable but stuck in its own head. It can only work with the information it was trained on, which isn't always up-to-date or connected to the real-time world.

Before MCP came along, if you wanted your AI to do something that required outside information or using another tool – like checking the current stock prices, accessing a document in your cloud storage, or sending a message through a different app – it was often a complicated process.

Developers had to build custom connections for each specific tool and data source. This wasn't just a headache for the technical teams implementing AI; it limited what AI could actually do for everyday users. The challenge before MCP can be understood as the "M×N problem".

If you have multiple AI applications (M) and numerous tools or systems (N) you want them to interact with, you potentially need M multiplied by N individual integrations. This leads to a lot of duplicated effort and makes scaling AI capabilities a real pain.  

The Gartner Hype Cycle

Introducing MCP

Enter the Model Context Protocol (MCP). Think of it as a universal translator or a standardised set of rules that allows different AI models to communicate and connect with a vast array of external tools and data sources in a much simpler way.

It's like establishing a common language that everyone in the AI world can understand, making it much easier for AI to access the information and tools it needs to be truly helpful. Anthropic, the folks behind the AI assistant Claude, introduced MCP as an open standard.

Think of an open standard like the rules of the road – everyone agrees to follow them, which makes things run much more smoothly for everyone. By making MCP open, they're encouraging the entire AI community to adopt this common way of connecting things. The fact that MCP is an open standard is a significant factor in its potential for widespread adoption. Unlike proprietary systems, an open standard allows for collaboration and innovation across different companies and developers, preventing any single entity from controlling the technology and potentially hindering its growth. 

USB-C for AI

Remember the days of having a different charger for every gadget? USB-C is a game-changer because it's so versatile. But the widespread adoption and "universality" of USB-C didn't happen overnight. It was driven by its inherent advantages – versatility, user-friendliness (reversibility), and the backing of major players in the tech industry, along with regulatory pushes like the EU mandate. 

Here's where the magic happens: MCP is aiming to be the USB-C of the AI world!

Just like USB-C provides a standard way to connect your devices to various peripherals, MCP wants to be the go-to, standardised way for AI models to connect with all sorts of "AI peripherals" – which in this case are those external data sources (like your documents in the cloud or information from databases) and tools (like APIs that let AI send emails or fetch specific data from the internet).

The idea is that instead of having a unique, complicated connection method for every single AI tool and data source, MCP offers a single, consistent protocol. So, if an AI model and a tool both speak "MCP," they can easily understand and work with each other, no matter who created them. This move towards standardisation simplifies the entire process of building AI applications that need to interact with the outside world. It means developers can spend less time wrestling with complex integrations and more time focusing on creating innovative and useful AI features. This mirrors the advantages of standardised interfaces in software development. 

No More AI Islands!

"MCP: Your AI's universal connection point, just like USB-C for your devices."

Think of AI models before MCP like isolated islands. They're powerful within their own boundaries, but they can't easily reach out and interact with the vast ocean of information and tools that exist beyond them.

This means they're often limited to the knowledge they were initially trained on, which can quickly become outdated in our fast-paced world. Imagine asking an AI about the latest news, and it only knows what happened before its last training update! This isolation problem often forces users into a frustrating loop – the "copy and paste tango".

You have to manually find the information your AI needs from various sources and then copy-paste it into the AI's interface. It's like having a super-smart assistant who can't open a door and needs you to hand them everything they need through a tiny window!  

MCP acts like a network of bridges connecting these AI islands to the mainland – a vast network of "servers" that provide access to all sorts of capabilities.

These servers can be anything from accessing files on your computer to interacting with online services like Google Drive, Slack, or even more specialised tools for coding or data analysis.

For example, imagine you're planning a trip. With MCP, your AI assistant could connect to a calendar server to check your availability, a flight booking server to find the best deals, and a maps server to figure out directions – all without you having to manually provide all that information!

MCP follows a client-server architecture with hosts (AI applications), clients (within the hosts), and servers (providing the capabilities). 

Why This Matters to You

The biggest win for you, the everyday AI user, is the sheer simplicity that MCP brings to the table. You no longer need a PhD in computer science to get your AI to play nicely with other services.

It's like going from needing a specific adapter for every single electronic device to just using a standard USB-C cable. Instead of developers having to build custom connections for every AI and every tool, MCP provides a more "plug and play" approach. 

Because AI models can now access the information they need more easily and efficiently through MCP, you'll likely experience AI interactions that feel much smoother and more intuitive.

Think about asking your AI to summarise a document from your cloud storage and then email it to a colleague. With MCP, this multi-step process becomes much more streamlined because the AI can directly access the document and then use an email tool, all through a standardised protocol. This leads to AI that feels less like a separate entity and more like an integrated part of your digital life, capable of handling complex tasks more autonomously. 

Taking the "Tech" Out of Integration

The beauty of MCP is that it handles the complex technical details behind the scenes.You don't need to worry about APIs, authentication, or data formats – MCP takes care of that.

This allows you to focus on what you want to achieve with AI, rather than getting bogged down in technicalities. Think of it like using a smartphone app – you don't need to understand the underlying code to use its features. MCP aims to bring that same level of ease to AI integrations, making AI more accessible to everyone, not just developers. 

The Future is Connected

So, what's the bottom line with all this MCP buzz? It boils down to connection – making it easier for our increasingly intelligent AI models to connect with the vast world of data and tools out there. Just like USB-C simplified how our gadgets interact, MCP is poised to do the same for AI.

For you, the everyday user, this means less fuss, more functionality, and AI experiences that feel more natural and helpful. You won't need to be a tech wizard to get your AI assistant to work with your favourite apps and services. MCP handles the behind-the-scenes complexities so you can focus on the amazing things AI can help you achieve.

The potential impact of MCP is huge. It could pave the way for a future where AI is seamlessly woven into the fabric of our digital lives, assisting us with tasks, providing us with real-time information, and making our interactions with technology more intuitive than ever before.

Here at TYMO AI, we're excited to see how MCP evolves and continues to make the world of AI more accessible to everyone. Stay tuned for more updates and explanations as this technology continues to develop!