Cognitive Exhaust Fumes, or: Read-Only AI Is Underrated
summary
A personal intelligence system that only reads six sources (email journal tasks CRM browser sessions notes) and never writes back is described with concrete tooling and outcomes: Python scripts call Anthropic/Claude, outputs land in an Obsidian vault, Vivaldi SQLite and CRM joins surface cross-source signals that reveal intention action gaps, attention drift, and relationship decay. The claim is that read-only constraint yields safer analysis not brittle automation the agent saves ~30 seconds but the observer flags a two‑week avoidance and suggests specific people to contact via mapped articles; repo provided.
AI-written summary · about
chapters (5)
related talks
42mAgents of Change: From Time Saved to Intelligence Activated
47mHuman Robot Agent: Scouting AI Beyond Hype and Doom
48mStop Fighting Your AI Engineering Prompts That Actually Work
46mIntelligence as Infrastructure: What Developers Must Learn to Build with AI
40mTurn 10,994 Notes Into Memory
38mThe Cognitive Cost of AI-Assisted Open Source
32mBeyond the Hype: A Pragmatic Look at How Developers Actually Use AI Tools
01h 00mKenote: AI Didn’t Replace You. It Promoted You!