Hilary Gridley runs her calendar, tasks, and daily routines by narrating to a Claude Code terminal. Teresa Torres types '/today' each morning and Claude scans hundreds of markdown files to generate her to-do list. Greg Isenberg uses Obsidian plus Claude Code to surface product ideas and automate research. Dan Shipper called Claude Code 'the most underrated AI tool for non-technical people.' Lenny Rachitsky collected 50 examples of people doing this. The results are real. The workflows are impressive. And that is exactly the problem.
Every single showcase, no matter how sophisticated, is about doing existing work faster. Torres's '/today' command does not invent new categories of thinking. Isenberg's second brain surfaces connections he might have made anyway. The author runs their own setup connecting Claude Code to GitHub, Obsidian, and dozens of AI agents, and calls it the most significant shift in how they work since Gmail. But here is what that experience taught them: Claude Code is a terrible interface. Janky, unreliable, hard to diagnose when it breaks. And the people celebrating it have built three-layer context systems and vault management philosophies, which means these tools do not meet you where you are. They reward you for already thinking like an engineer.
The piece's sharpest argument is not about UI quality. It is about exclusion. People who already think in workflows, decision trees, and structured data are being elevated enormously right now. Everyone else is being left behind, and faster than anyone is admitting. The goal the author lays out is not making systems thinkers more productive. It is unlocking people who think creatively, associatively, and intuitively. That argument, and the data behind it, is what makes this worth reading in full.
[READ ORIGINAL →]