A hybrid, local-first system that bypasses cloud IP blocking to scrape, synthesize, and pipe YouTube transcript data into your personal knowledge base with $0 server costs.
YouTube aggressively limits cloud hosting providers. To solve this, I engineered a hybrid execution model: ingestion sits in the cloud, while scraping executes on a local residential IP node, simulating authentic traffic patterns.
Lightweight local nodes running in Docker containers subscribe to task queues via Webhooks. This ensures reliability without the overhead of massive VM clusters.
CF Worker captures PubSub webhook.
Extension prompts human-in-the-loop validation.
Residential IP node executes secure extraction.
Gemini AI synthesizes and categorizes transcript.
Markdown files pushed to Obsidian/Telegram.
Zero-latency webhook listeners on the edge, handling thousands of concurrent channel updates.
A custom dashboard to manage your ingestion queue directly from the browser window.
Dynamic prompt templates that adapt based on the video niche (Tech, Philosophy, Gaming).
Utilizing free-tier cloud resources and existing local hardware for 100% margin operation.
Automatic generation of clean Markdown files with bi-directional links for your PKM.
Daily summaries of processed insights delivered as a clean message to your mobile device.