From the AI Proving Ground to My Daily Workflow: How MCP Powers My Day-to-Day AI Tools
In this blog
This post is part of our Everyday AI series crafted by WWT AI experts to enhance awareness and comfort with Generative AI (GenAI). Our goal is to empower you to harness GenAI's diverse capabilities and benefits, both professionally and personally.
At World Wide Technology, we spend a lot of time experimenting with different AI tools and products, and we have quickly learned that there's no shortage of AI tools available today. From chat interfaces to coding assistants and agent frameworks, the ecosystem is exploding with options for different AI tools that can benefit a wide range of people. However, what happens if you step outside of the data center and the capabilities we have access to in the AI Proving Ground? The question becomes:
If you're anything like me, when attempting to implement these AI tools into your life, you will quickly hit your first wall which is setup. Before using all of these AI tools, you have to go through quite a bit of setup and integration to make them actually functional, and that friction is often enough to stop momentum altogether.
MCP didn't make AI smarter — it made it usable
This is where I decided to lean into the Model Context Protocol (MCP), and more specifically, the Docker MCP Toolkit.
MCP is an open-source standard that defines how AI applications (called MCP clients) discover and interact with external tools, data and systems. Those capabilities are made available through small, dedicated programs called MCP servers, which expose specific tools and resources in a structured, consistent way that AI clients can understand and use. In simple terms, MCP creates a consistent way for AI clients (think ChatGPT or Claude Desktop) to find capabilities, understand what they do and call them.
What I am really leveraging MCP for is to give my AI clients, for now its LM Studio, access to external tools. These external tools will allow my AI client to move beyond the chat window and start interacting with things like my notes or local files. The other benefit of using MCP is one day if I need or want to change AI clients, I can do so and still have access to my external tools since MCP was designed to be a universal standard.
What changed everything for me was Docker's implementation. The Docker MCP Toolkit lets you run MCP servers as containers directly on your machine. Instead of building everything from scratch, I could choose from a growing ecosystem of prebuilt MCP servers, spin them up like any other container, and immediately start using those tools from within my AI clients.
That moment when tools simply appeared and worked is when MCP turned implementing these AI tools into my daily workflow from a monumental task into something that seemed within reach.
Using MCP in my everyday life: Obsidian for notes and weekly planning
One of the first MCP servers I added was the Obsidian MCP server, mainly because I have recently started experimenting with Obsidian in my daily workflow as a notes taking application after seeing others use it, and I wanted to see how it would work for me. I use Obsidian to keep track of ideas ideas, create weekly to-do lists and just overall create notes for myself. It made sense to see if AI could make my use of Obsidian even more efficient. With Docker MCP Toolkit, getting the Obsidian MCP server stood up took only a few clicks, and it was ready to be used.
With the Obsidian MCP server running, my AI clients can interact directly with my notes and that's quickly changing how I use Obsidian. Instead of opening a note and manually writing everything down, I can ask AI to quickly jot down notes as I talk through an idea, summarize scattered thoughts into something more structured, or even generate a simple weekly task list based on what's in my vault.
For instance, I asked LM Studio:
After briefly thinking of what tool to use and the content of the note, I am notified that the note has been successfully created and a small summary of what it contains is given. When I then go to my Obsidian notes, I see the following:
It's a small shift, but it adds up. Capturing ideas feels lighter, organizing them feels easier, and because the MCP server connects AI directly to my notes, I don't have to constantly copy, paste or reformat things between tools. After having a good experience with the Obsidian MCP server it got me thinking what else could I do through MCP?
From prebuilt MCP servers to building my own
After spending some time working with prebuilt MCP servers, I started to wonder how far this idea could really go. If MCP servers are just small programs that expose tools to AI clients, what would it look like to build one around something I actually wanted in my day-to-day life?
That curiosity led me to try building a simple custom MCP server of my own: a reminder service. The goal was straightforward: give my AI client the ability to create, list and delete reminders on my Windows machine, and surface them as native toast notifications. After building out the necessary files and running the MCP server as a docker container, it was once again time to test out my new tool. I requested a simple reminder to be triggered in five minutes:
After the five minutes, I received the reminder on my windows machine:
Building that small server completely changed how I looked at AI clients and MCP. It made it clear that MCP servers don't have to be massive platforms or enterprise integrations. They could be small purpose-built tools that if I put in the time and effort to create would allow me to leverage AI more than I ever thought possible.
Still early, and that's the exciting part
Docker MCP Toolkit and MCP have already changed how I think about using AI in my day-to-day life. They didn't make AI smarter. They made it easier to connect AI to the tools, notes and systems I already use. That shift alone has made AI feel far more practical outside of controlled lab environments.
That said, I'm still very early in this journey. There's a lot more to learn about MCP, a lot more servers to explore and a lot more ideas to test. What feels different now is that experimenting doesn't come with the same overhead. Trying something new feels approachable.
If you have any interest in MCP or experimenting with the Docker MCP Toolkit, check out the following lab coming out of the AI Proving Ground: Model Context Protocol (MCP) Foundations Lab.