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.
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:
Extracts atomic ideas from your messy, rambling notes
Clusters related concepts using semantic embeddings
Distills and deduplicates overlapping thoughts
Creates clean, structured knowledge with backlinks to sources
The system provides a clear visualization of the transformation:
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:
Source Component: Uses LlamaIndex to read raw markdown files from Obsidian
Extractor: Identifies and separates atomic ideas from raw notes
Embeddings: Creates semantic vector representations using OpenAI's embedding model
Clustering: Groups related atomic ideas together
Distiller: Merges and deduplicates clusters into coherent concept notes
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:
Milestone 2:
Scaling performance improvements
Building a user interface for file uploads
Processing entire vaults, not just inboxes
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: