How you can create your second brain with Claude

The traditional way we interact with artificial intelligence is transactional: you open a chat, paste a task, get an answer, and close the tab. The next time you log on, you start from zero, re-explaining your preferences, your business, and your goals all over again.

Unlike your best friend or colleague, Claude doesn’t learn about you from all the conversations it has had with you.  

Hence, if you want Claude to carry on from where the discussion was halted at the moment, you likely need to set a context with your first prompt. But what if Claude could learn about you from your discussions?

The second Claude-powered brain and why people are creating it

Popularised by tech leaders like Andrej Karpathy under the “LLM Wiki” pattern, having a second brain using Claude or any other LLM model helps in having a smart experience. You essentially create a unified system inside a chatbot where you have a persistent and local knowledge base – acting as a storage for everything you have discussed with the LLM. In a second brain, Claude acts as your cognitive layer that reads, connects, and expands on everything you discussed with the chatbot. With every new conversation, you stop re-explaining yourself to your AI chatbot because the system already remembers and knows you.

What’s more fascinating is that unlike a human brain, setting up your second brain up takes less than an evening. Moreover, this mind gets sharper every time you use it. 

Did we get you curious? Here’s how you can build it, using an easy technical guide shared by Yarchi, an AI researcher, who goes by @undefinedKi on X (formerly Twitter).

Before we start, though, here are a few clarifications:

What all do you need to build a second brain

To create your second brain, you need more than just Claude. Remember, Claude only works as the cognitive layer that does the thinking – we need to create a storage for everything Claude can work on. We need to create a memory for your brain, which Claude can read. Hence, let’s look at the two things you need to create a digital brain.

1. Obsidian – the ‘memory’ for your brain: 

Obsidian is a free notes app that stores everything in plain-text Markdown file format on your computer. Notes created on Obsidian can link to each other, forming an interactive graph view over time. Since these are stored on your local hardware instead of cloud storage, these files remain easy to access and you own them forever.

2. Claude – the cognitive part of your brain: 

We will use Claude Desktop’s native terminal power, since Claude will read your entire vault, file new information where it belongs, link related concepts, and accurately answer queries across your entire usage history.

With that out of the way, let’s look at how you can create your second brain.

Step 1: Install Claude Desktop

You will need to download the Claude desktop app from the official page, regardless of whether you want to use it for Windows or Mac. Sign in with your Claude account and note that you will need a paid plan (such as Claude Pro) to access advanced execution features like Claude Code or agentic workspaces.

Locate the advanced tabs inside the app, looking for the Code or terminal execution tab. This allows Claude to interact natively with your computer’s filesystem.

Step 2: Establish your Obsidian Vault

Next, you need to download Obsidian – the free note-taking app that will store information in the ‘.md’ format on your computer – from its official website. Once installed:

– Open Obsidian and click “Create new vault”.

– Give it a name, for example, a brain. Then choose a secure folder on your local drive to store it.

– Create a test note to understand the core mechanic. Type a sentence, then type two square brackets around a word, like [[goals]]. This creates a bidirectional link. Over time, Claude will automatically generate these links for you, building an organic map of your thoughts.

Step 3: Setting up Obsidian

To allow Claude to access your notes, Obsidian needs to safely open a local communication port. We do this via an open-source community plugin.

– In Obsidian, click the Settings (gear icon) in the bottom-left corner.

– Navigate to Community plugins and select “Turn on community plugins”.

– Click Browse, search for “Local REST API”, click install, and then toggle it to Enable.

– Open the Local REST API plugin settings. You will see an API Key (a long string of letters and numbers). Copy this key, since you will need it in the next step. Keep Obsidian running in the background.

Step 4: Wire the connection with MCP

Next, you need to command Claude to connect to that open port. Navigate to your Claude Code/terminal interface inside Claude Desktop and execute the following configuration command, making sure to swap ‘PASTE-YOUR-KEY-HERE’ with your actual Obsidian API key you copied earlier:

claude mcp add-json obsidian-vault ‘{

  “type”: “stdio”,

  “command”: “uvx”,

  “args”: [“mcp-obsidian”],

  “env”: {

    “OBSIDIAN_API_KEY”: “PASTE-YOUR-KEY-HERE”,

    “OBSIDIAN_HOST”: “127.0.0.1”,

    “OBSIDIAN_PORT”: “27124”

  }

}’

(Note: When pasting your key, ensure you do not copy the word “Bearer” if it appears in the Obsidian UI—only copy the alphanumeric string).

To test whether it worked, type: “List every file in my Obsidian vault.” If Claude prints your notes back to you, the system is officially live.

Step 5: Create the master profile

You now have an empty brain, which needs to be filled with useful directions. Instead of manually writing out your life story, we will force Claude to interview you. In your Claude chat window, paste this prompt:

“You are setting up my second brain. Interview me ONE question at a time to build my profile. Ask about: who I am and what I do, my goals for this year, how I want you to talk to me, my strengths and weaknesses, and my current projects. Wait for each answer before the next question. When finished, write everything into a file called CLAUDE.md at the vault root, structured with headers, so you load it automatically every session.”

The more honest and detailed your answers, the sharper your second brain becomes. Once the interview concludes, a CLAUDE.md file will appear at your vault root, preserving your global context forever.

Step 6: Build your project pipelines

Your root ‘CLAUDE.md’ establishes your high-level strategy, but daily work requires hyper-focus. You need to segment your life into individual project directories so Claude isn’t overwhelmed by irrelevant data. Hence, run the following prompt:

“Create a project folder in my vault called youtube-channel. Inside it, create four folders: Inputs, Process, Outputs, and Feedback. Then write a CLAUDE.md inside that project folder describing what this project is, its primary goal, and your specific role in helping me hit it. Interview me if you need details.”

You can swap youtube-channel for whatever you are actively building. You now have a productivity pipeline, i.e., raw thoughts go to Inputs, Claude structures them in Process, finalised work gets built in Outputs, and performance metrics land in Feedback.

Step 7: Isolate context for execution

In Obsidian, select your vault name Manage vaults >  Open folder as a vault  >  pick your project folder  > Trust. 

When you open Claude inside this isolated directory, it reads only that project’s dedicated CLAUDE.md. It isolates its attention entirely on the task at hand, minimising token hallucination and maximising output precision.

Step 8: Standardise workflows into “Skills”

If you find yourself executing a task more than once, such as drafting social medua posts, formatting client reports, or reviewing code, turn it into a permanent skill file. Inside your project, tell Claude:

“I want to turn this into a reusable skill. Here is how I perform [insert your task], alongside an example: [paste your template or example]. Save this as a markdown skill file inside this project’s skills folder, featuring a clear name and instructions on when to trigger it.”

Now, you can simply command Claude to “run the [Name] skill,” and it will execute the workflow automatically using your exact specifications.

Step 9: Integrate live workspace data

A second brain shouldn’t just rely on manual notes, hence it should hook into real-time variables. You can easily connect your Google Workspace, Slack, or email via the MCP terminal. For instance, to connect Google Calendar, run:

claude mcp add google-workspace uvx workspace-mcp –tools calendar

Complete the secure OAuth sign-in window that pops up. Once linked, you can hand off agentic workflows seamlessly: 

“Review my calendar for today, log what I committed to across my meetings directly into my tasks folder, and flag any item lacking a clear next step.”

Step 10: Put the ‘Brain’ on autopilot

Once your skills are operational, use Claude’s built-in scheduling or automation capabilities to run routine upkeep. You can navigate to the scheduling interface or issue a recurring directive:

Frequency: Daily, 7:00 AM

Folder: Your local vault

Prompt: Review my vault. Sort any raw notes sitting in my Inputs folder into their respective directories and link them to related topics. Flag any notes that haven’t been updated in 30 days. Write a concise, 3-line summary of what was organised overnight.

Every morning, you will wake up to a knowledge base that has neatly indexed, sorted, and filed itself while you slept.

You have now created your digital brain

With a Claude-powered digital brain, you bypass standard, short-lived chat windows, wherein you are no longer prompting a generic chatbot. 

Instead, you are constructing a personalised digital extension of your mind. Granted, it is a bit technical and complex to set up, but you now have an intelligence layer in Claude that picks up precisely where you left off, and builds a relational knowledge graph that expands and connects your ideas automatically, just like a human brain. In essence, Claude knows who you are and what you are referring to.