Recommended: be more ambitious with models

Felix Rieseberg, who leads engineering for Claude.ai, Cowork and Claude Code Desktop at Anthropic, wrote How to be more ambitious: a tutorial for humans with models, and it's the kind of post I want to reread in a few months. The idea that sticks is simple: the limit is rarely the model, it's what we believe we can ask for. Using AI only to do the same work faster is the lowest form of ambition. He suggests going up in altitude (asking for the outcome behind the task, not the task), going deep (working the way the best expert in the field would, with raw data and an outside judge checking the result) and widening the range (bringing in methods from fields the problem wouldn't normally call on). It's here so I remember next time I ask for too little.