Self-Organizing Second Brain

How I Manage Information Overload

I’m drowning in information. Gen-AI makes it worse.

As someone who relies heavily on my second brain system, I've hit a wall.

The more notes I take, the less useful my system becomes. My Obsidian vault has turned into a cluttered graveyard of valuable thoughts rather than the dynamic thinking tool it was meant to be.

I just can’t keep up with my inbox.

If you're drowning in notes and spending more time managing your personal knowledge management (PKM) system than actually using it, you're not alone. This problem led me to build something I believe many of us need: a self-organizing second brain.

Here’s OpenAugI, an AI system for organizing and distilling my second brain.

My Inbox - extracted into atomic notes, embedded into 2D

The Problem: Our Brains Don't Work Like Filing Cabinets

Traditional note-taking advice tells us to be better curators—to tag more carefully, build complex organization systems, and essentially become better librarians of our own knowledge.

I don’t want to be a librarian, I want to work.

And here's the fundamental problem: our minds don't work sequentially. We think in multiple parallel threads, with our subconscious processing various ideas simultaneously. Traditional organization systems simply don't match our natural thought patterns.

My Obsidian inbox alone contains 274 notes. Capture isn't the issue—one-tap to Obsidian from my phone, voice notes, and Readwise ensure I never lose an idea. But processing these notes has become impossible with my limited time.

The result? I repeat myself, lose track of previous thoughts, and struggle to build upon earlier insights. The cognitive overhead of organization has become a bottleneck to actual thinking.

The Solution: OpenAugI

Instead of generating more content (the last thing we need), I built an augmented intelligence system focused on distillation and organization.

OpenAugI is an open-source knowledge distillation framework that:

1. Extracts atomic ideas from your messy, rambling notes

2. Clusters related concepts using semantic embeddings

3. Distills and deduplicates overlapping thoughts

4. Creates clean, structured knowledge with backlinks to sources

The system provides a clear visualization of the transformation:

Distilled “concept notes” overlayed onto raw atomic notes.

In the visualization above, each blue dot represents an atomic idea extracted from my raw notes. The larger points show the distilled concepts that emerged after clustering and merging related ideas.

How It Works: The Technical Pipeline

OpenAugI uses a modular pipeline architecture similar to frameworks like LangChain and txtAI:

1. Source Component: Uses LlamaIndex to read raw markdown files from Obsidian

2. Extractor: Identifies and separates atomic ideas from raw notes

3. Embeddings: Creates semantic vector representations using OpenAI's embedding model

4. Clustering: Groups related atomic ideas together

5. Distiller: Merges and deduplicates clusters into coherent concept notes

6. Knowledge Store: Saves everything in LanceDB with metadata and backlinks

When I ran this process on my 300-note inbox, the entire pipeline cost about 30 cents using OpenAI's embeddings and a small LLM for distillation.

From Chaos to Clarity

What makes this approach powerful is how it aligns with our natural thinking patterns. Instead of forcing you to organize hierarchically, it allows ideas to cluster organically based on semantic similarity.

For example, one cluster in my notes contained atomic ideas related to "management organization systems" and "implementing a distillation service." These previously scattered thoughts were automatically grouped and distilled into a single coherent concept.

This means I can capture freely, knowing the system will:

• Break down my notes into atomic components

• Route each idea to its related thought cluster

• Build upon previous thinking without manual organization

• Preserve context through backlinks to source notes

The Roadmap: Just Getting Started

This prototype represents just the first milestone in building a complete augmented intelligence system. The roadmap ahead includes:

1. Milestone 2:

   • Scaling performance improvements

   • Building a user interface for file uploads

   • Processing entire vaults, not just inboxes

2. Milestone 3:

   • Creating a human-in-the-loop workflow

   • Implementing a pull-request-like interface for reviewing changes

   • Allowing editing of suggested distilled notes before acceptance

Adapting to the Intelligence Age

The future isn't about AI replacing human thought—it's about augmenting it. While most AI tools focus on generating more content, OpenAugI focuses on helping you make sense of what you already have.

My vision is to merge the best of what AI can do with the best of what human brains can do.

By offloading the cognitive burden of organization to AI, we free ourselves to think, create, and build.

Help Build the Future of Augmented Intelligence

If you're drowning in notes and struggling to make sense of everything you capture, I want to hear from you. I'm looking for alpha testers who can help shape the future of OpenAugI.

Visit openaugi.com to join the waitlist and get early access to the prototype.

Head to GitHub to join the conversation.

Remember: Augment, Stay Human.

If you prefer video: