Escaping “Information Rot”
“I watched an incredible 40-minute lecture on YouTube, sent the link to a friend, and then… I forgot everything. I realized that saving a link is not the same as capturing knowledge. That insight was lost in a graveyard of ‘Watch Later’ lists.”
We are drowning in more high-quality information than any generation in history, yet we remember less of it. I created Nexus to stop “Information Rot”—the slow decay of valuable ideas that are captured but never processed.
🏗️ The Architecture: “The Cloud-Inbox”
Nexus uses a unique “Local-First” architecture. It treats the cloud as a high-powered, zero-cost “Inbox” while keeping your private data on your own machine.
1. The Cloud: A Private Post Office
When you find a link on your iPhone, Nexus wakes up in GCP Cloud Run. It “watches” the video, “reads” the transcript, and synthesizes Atomic Notes. It stores these notes in a temporary “GCP Inbox” until your Mac pulls them in.
2. The Index: Privacy-Aware Linking
To help the AI link new notes to your existing knowledge, your Mac “pushes” only your Note Titles and Aliases to GCP Firestore. The cloud never sees the content of your notes—only a “Map” of your vault.
3. The Mac: The RAG Engine
Because your notes stay local, you can use high-powered CLI tools to talk to your entire vault. You can ask: “Based on my last 6 months of research, what are the recurring patterns in my thinking?” and get instant, private answers grounded in your own unique knowledge base.
🛠️ The Engineering: Mastering Serverless
To build this for $0/month, I had to solve major technical hurdles that separate a “toy” from a production-ready system:
- Scale-to-Zero: Built on GCP Cloud Run, the system costs $0 until you send a URL. It wakes up in seconds, processes the request, and disappears.
- Bypassing Blocks: I implemented a specialized environment using the Deno JS runtime and PO Tokens to bypass YouTube’s data center blocks, ensuring 100% reliable capture where standard scrapers fail.
- Persistent Index: Since serverless containers forget everything, I moved the “Map” of the vault to Firestore, ensuring the AI always knows how to link new notes.
🚀 The Result: A Universal Knowledge Pipe
Today, Nexus is my Universal Inbox for all digital information:
- YouTube/Instagram: Extracts core arguments from video and audio using
yt-dlpandffmpeg. - Web Articles: Uses Jina Reader to “distill” long-form articles into atomic insights.
- Research Papers: Parses PDFs to extract methodology, results, and key takeaways.
Because of the Semantic Search I built using Firestore’s vector queries, I frequently experience “Aha!” moments where the system reminds me of a connection I would have never made manually. My vault is no longer a graveyard of links; it is a Living Knowledge Base that grows more valuable with every capture.
🧠 Obsidian Workflow
My “Second Brain” is built on the Zettelkasten method, utilizing:
- Atomic Notes: Single-concept notes for maximum linkability.
- Local RAG: Using private models to query the vault without data leaks.
- MOCs (Maps of Content): Curated hubs that organize evolving thoughts.
- Permanent Index: Ensuring every new capture from Nexus finds its logical home.