Understanding what MCP means - OAuth Tutorial

- Now, we have many protocols already in existence that we use all the time. The one that you use all the time, that you are using right now if you're watching this course or watching the livestream, is the Hypertext Transfer Protocol or HTTP. And the Model Context Protocol is another type of protocol. It is not software that runs anywhere, it is a way for systems to talk to one another. MCP was released in November last year to not a lot of fanfare. It popped up as a blog post on Anthropic's blog, Anthropic is a company that makes Claude. And developers looked at it and went, "Interesting." The reason why is because of the description. The Model Context Protocol is an open standard that enables developers to build secure, two-way connections between their data sources and AI-powered tools. The architecture is straightforward. Developers can either expose their data through MCP servers or build AI applications, MCP clients, that connect to these servers. The name, Model Context Protocol, from which MCP stems, actually describes what it is, it's just backwards. It is a protocol that provides context to language models. A protocol are the formal rules that computer systems use to communicate with each other. I already mentioned HTTP as a protocol. There are other protocols as well, we have FTP, SMTP, SMS, like all these different protocols that we use. Context is the information that is sent to the AI system to work with, and the model is the AI system itself. Now, if you've ever worked with Claude or ChatGPT or Gemini or any of the other tools that we currently have, you know that to get the systems to do anything meaningful, you first have to provide context to them. In the most rudimentary way, that means going to a document or to a spreadsheet or to a webpage or somewhere else, copying a bunch of content out, then pasting it into the AI chat, and say, "Hey, help me work on this." Or, "Analyze this information." Or, "Help me rewrite it." Or, "Find information from within it." Or something like that. That is the context. It also illustrates how difficult it is to get AI to actually be helpful, because on its own, these conversational AI tools, like ChatGPT and Claude, can only work within their own context, so only within the conversation that you're having. And that means the things that we imagine these tools should be able to do, like go on the internet and find information about something or clean up my inbox or clean up the spreadsheet or something like that isn't as straightforward as it seems, because the chat is just the chat, and any external connection needs to be made manually.