Lazy slobs globally are getting super excited about MCP, and here is why…

Anthropic released the MCP back in late November 2024, and since then, experienced and inexperienced developers and enthusiasts have begun to really show what is possible with this new open standard.
‍‍

What is MCP?

MCP stands for Model Context Protocol, an open-source communication standard developed by Anthropic that connects AI assistants to applications and databases.

So, what does that mean to smooth brains like you and me?

Well, on their own, LLMs like ChatGPT, Claude, GROK, etc., are great. They have drastically increased global productivity and are well on their way to replacing how we browse the Internet today.

To get to the promised land of having specific agents and assistants who know you better than you know yourself, helping you in everyday mundane tasks like:

“Summarize all my emails, Telegram messages, Discord chats, and Slack notifications and then automate my responses and tasks to complete them.”

This near-future sounds like a beautiful dream. I am the laziest person on the planet and find these repetitive tasks take up a lot of my time, energy, and bandwidth.

Traditionally, I would hire a personal assistant to do them, but then I would have to spend money and time training them on how I want these tasks to be completed each day, which, in all honesty, just creates more work.

And besides, I don’t really want another person to have all my private and sensitive information. Now, this slowly puts us on the path of, well, do you really trust the LLM providers, the application developers, and the middle layer of MCP itself?

I don’t want to go down that road in this piece, but I like where we are headed with things like zkTLS to circumnavigate this — only share what is absolutely necessary, and this can be verified.  

Anyway, the issue with my pipe dream mentioned above is that the vast variety of LLMs have no standard connection/communication protocol to be able to connect to the even greater number of applications and databases that we all use each and every day.

This is where MCP comes in. It serves as the standard for connecting AI to applications in two-way communication.

This is a 10-100x unlock in productivity for technical and non-technical roles. Every company will begin to build MCP connections to their products both internally for their staff and externally for their users.

There is no reason that Sky Scanner would not be one of the first to try to implement this MCP connection for existing LLMs or even their own AI travel agent.

Travel is the cliche way to show what LLM-to-agent interactions would look like, but it is worth using it again to highlight the point here. Even OpenAI’s Operator demo used this flight booking example to highlight the power of agent-to-app cases.

MCP and its potential in crypto

Blockchains are extremely data-rich. However, even the best block explorers and analytics platforms leave a lot for the end user to interpret.

Tools like Dune and Flipside have helped. Anyone can query data, but again, this requires a very strong understanding of SQL and how to read a block explorer and its many contracts.

A few developers are working on connecting LLMs directly to block explorers to enable users to query on-chain data in real-time. This would level the playing field for those wanting to access this data but don’t want to rely on third-party platforms and analysts to create the data and dashboard for them.

‍
The future of on-chain analytics is in the hands of those who know what to prompt for. We are edging ever closer to a total level playing field, and those with a deeper than surface-level understanding of the underlying protocols and applications will have the upper hand, and this will snowball.

The gap between those in the know and those who do not have a clue will increase drastically. If you know how to access these tools and how to prompt for the correct data, then you will do extremely well while the rest of the market is figuring out how to catch up.

A lot of the top DeFAI projects, like Griffain, Orbit, HeyAnon, etc., are almost certainly playing this out currently, which is majorly bullish, IMO. I am less excited about “go swap me $100 SOL into the most amount of USDC I can get.”

But even more so excited about: “Go create a report using Deep Research and query the latest Dune, Tokenomist, and blocmates articles to decide the top ten protocols with the highest revenue per market cap.”

“Then, provide me with solutions to allocate $1,000,000 across them all. Make sure to report on potential upcoming catalysts, unlocks, and anything that affects supply for the better or worse.”

“When you are done, go use Kaito to check the social sentiment around each project and rank them accordingly.”

“After that, go through the blocmates Discord and my personal chats to highlight my inner circle's opinions on these projects and include them in each relevant section.”

“Once you are happy, suggest the portfolio allocation, the reasoning, and the invalidation (price and fundamentally) to me, and then find the best route to purchase said tokens.”

Bottom line

So, yeah, MCP is a huge unlock for productivity, period.

We are now entering a world where scaling through employees is less and less likely, for better or worse. Scaling through a fundamentally strong understanding of the industry you are in and having taste, understanding positioning, culture, and a go-to-market is now more important.

I’m super excited to see where this goes. If there are any teams building towards this vision, please let me know. I’d love to chat.

Signing off,

Grant