Clear Thinking is the Top 1% Skill in the Age of AI

There’s a chain reaction most people miss: Clear thinking → Clear writing → Clear prompting Each step refines the one before it. Writing forces you to confront the gaps in your thinking. Prompting forces you to confront the gaps in your writing. AI amplifies whatever you input. Feed it muddy thinking, get muddy outputs. Feed it precision, get leverage. The Surplus of Competence We’re drowning in tools that enable a base level of competence. AI is one of them. Anyone can generate “pretty good” now - content, code, strategy, whatever. ...

January 19, 2026 · 2 min · Austin Johnson

The AI-Augmented Learning Thesis

A conversation that started with a question about courses vs mentors vs AI, and ended somewhere unexpected. The Question That Started It Can you compare buying a course and following a traditional top-down program versus working directly with a mentor—and more specifically, compare that to using AI as that mentor? Part 1: What the Research Actually Says Bloom’s 2 Sigma Problem (1984) Students who received one-on-one tutoring performed two standard deviations better than students in conventional classrooms. The average tutored student outperformed roughly 98% of students learning in a traditional class setting. ...

January 18, 2026 · 8 min · Austin Johnson

Why You Can't Prompt Your Way Out of Not Thinking

LLMs are amplifiers. They amplify your thinking. They amplify your prompts. They amplify your direction. But here’s what nobody tells you: if there’s nothing to amplify, you get nothing useful back. The Problem You ask the AI a question. It gives you a fluent, confident answer. But you have no idea if it’s right. You can’t evaluate it. You can’t tell if it’s missing something crucial. You can’t catch the subtle errors mixed in with accurate information. ...

January 16, 2026 · 6 min · Austin Johnson

I Spent a Year Using Zettelkasten Wrong

For over a year, I followed zettelkasten principles religiously. My notes were atomic, containing one idea each. They were linked together with connections everywhere. They were beautifully formatted with clean markdown, headers, and structure. And I created them immediately while learning, exactly as I thought I was supposed to. My notes looked beautiful. They were technically correct. And they were completely useless. The result of all that work was hundreds of notes and zero value. Nothing to show for it. I would spend three hours on a thirty-minute course because I was creating “perfect zettelkasten notes” on every concept. By the time I finished, I was exhausted, the notes were pristine, and I never looked at them again. ...

January 12, 2026 · 14 min · Austin Johnson

Building a Personal AI Infrastructure

The AI tools you use daily should get smarter about you. I spent today installing what I’m calling Jeff - my Personal AI Infrastructure. It’s based on Daniel Miessler’s concept: instead of starting fresh every session, build persistent systems that learn your patterns, preferences, and workflows. The insight is simple. Every time you work with Claude, you’re generating signal. Session history, tool patterns, what works and what doesn’t. That signal usually disappears when you close the terminal. PAI captures it. ...

December 31, 2025 · 5 min · Austin Johnson