I used to think tools like ChatGPT or NotebookLM were the pinnacle of personal AI. I could talk to my notes, get contextual insights, and reflect on my work in ways that felt futuristic.
But then I hit a wall.
I wanted that same intelligence across all my tools, not just in a single platform. As I was trying to make connections and connect the dots across spaces, I realized the limitations of not just the Generative AI models, but also the ways in which I structured and used my work. I started thinking more about process and less about product.
That’s when I found something that changed how I work entirely:Model Context Protocol (MCP).
What Is MCP?
Model Context Protocol (MCP) is a method for giving AI the right information at the right time, based on what you’re working on.
Think of it like handing your AI assistant a custom folder of your most relevant notes before you ask a question. No guessing. No re-uploading. Just context-aware intelligence , embedded inside your existing tools.
Imagine you’re writing a paper. RAG models, like NotebookLM, are like building a Trapper Keeper. You organize your notes, articles, and worksheets in one place. When you want help, you open it up and point to the stuff the AI should look at. You can even say, “Use just these pages,” or “Mix in a little internet info too.”
It’s organized, but you still have to tell the AI where to look.
MCP is like giving your AI a backpack full of your stuff. Your notes, ideas, and schoolwork, so it always knows how to help you best.
Instead of flipping through five messy notebooks, your AI shows up with just what you need, already highlighted and cross-referenced. It can even go back into your notes, across sources and tools, and make edits to those if you want.
That’s MCP. And it’s not just an idea. It’s already working behind the scenes in tools today.
Why NotebookLM Wasn’t Enough
NotebookLM helped me “talk to my notes”, but with guardrails I couldn’t remove:
- I had to manually upload everything.
- It only worked inside one app.
- It didn’t scale as my notes grew.
- I couldn’t control what context the AI used or how.
I needed a more flexible, private, portable solution. Something that could travel with me across tools like Obsidian, Google Docs, VS Code, or wherever I work.
From Chatbot to Thinking Partner: What MCP Enables
Here’s what changes when you use MCP:
| Without MCP | With MCP |
|---|---|
| Generic AI stuck in one app | Personalized AI across your ecosystem |
| One-time uploads | Real-time, reusable context packs |
| Tool-dependent experiences | Tool-agnostic AI intelligence |
| Fluffy summaries | Deep insights grounded in your work |
Instead of adapting to the limitations of each AI app, you bring your context with you, and the AI responds as if it truly knows your thinking.
My Current Workflow with MCP (Simplified)
Here’s how I use MCP in my everyday knowledge work:
- Curate context packs in Obsidian
→ Tags and folders like “🧠 Digital Literacy,” “🎓 AI in Education,” and “📁 Workshop Materials.” - Use an MCP-enabled tool
→ I define which context pack is “active” when I’m writing, researching, or reflecting. - Ask precise, layered questions
→ “What themes connect my last 3 blog posts on AI bias?”
→ “What notes tie digital literacy to student agency in my workshop drafts?” - AI responds with clarity, not fluff
→ It doesn’t just summarize—it helps me think more clearly, connecting the dots across time and projects.
Do I Need to Be a Developer?
Not at all.
MCP is built for accessibility. Think of it like RSS for your second brain:
- You choose what gets pulled in.
- You define the scope.
- You stay in control of your knowledge flow.
And while early adopters are using open-source setups or tools like Claude MCP, more user-friendly options are coming fast.
Why This Matters for Educators, Researchers, and Creators
MCP isn’t just for techies. It’s for anyone working with complex ideas:
- Teachers can build AI-aware curriculum assistants tailored to their materials.
- Researchers can synthesize years of annotated readings across projects.
- Students can build their own “thinking partners” grounded in their learning logs.
- Writers & creators can remix their work with deeper clarity and continuity.
This is personal AI , on your terms.
Final Thought: AI That Understands You
It’s not about finding one perfect AI tool. It’s about using the right tools and knowing how they work together.
MCP is one possible bridge. It doesn’t replace your brain; it connects your notes, projects, and thinking across tools so AI can actually help you think.
It works best alongside your broader AI practice, where you consult multiple models, compare their perspectives, and synthesize insight.
So yes, we need more than one AI. But we also need better ways to help those AIs understand us , our work, our context, and our questions.
That’s what MCP makes possible.
Coming Soon : I’ll walk through how I build my context packs in Obsidian, and how you can start layering MCP into your own system without writing a single line of code.
Subscribe or follow to stay in the loop, and let’s bring your notes to life.