The Automation Lab

A collection of autonomous agents and high-leverage workflows built to distill signal from the noise. These systems are designed to scale awareness and capture intelligence in real-time.

The Philosophy

Information overload is fundamentally a bad design problem. I build these tools to bridge the gap between my brain's finite capacity and the internet's infinite firehose.

I design these pipelines as high-performance "noise-cancelers" for data. They utilize infrastructure like watermark deduping (to stop seeing the same thing twice) and quality gates (to make sure the output actually makes sense).

The real magic is in the plumbing. I've built "guru-voice" filters to kill the fluff, adversarial loops that double-check the facts, and a synthesis layer that connects dots across 60+ sources so I can stay informed without the existential dread of 500 open tabs.

Current Lab Stack

  • LLM: mimo-v2.5 / MiniMax-M2.7
  • Orchestration: Hermes Agent cron + subagent delegation
  • Data: SQLite watermark dedup + signal storage
  • Enrichment: Brave Search + Jina Reader
  • Delivery: Cloudflare Pages + Discord
  • Infra: macOS / Python 3.11