Running Out of AI: How Usage Limits Give Ideas Time to Develop
danshipper · x · 2026-09-14
Jack Cheng, senior editor at Every, reflects on how exhausting his Claude usage limits each week reshapes his workflow: he saves the hardest questions and one-shot builds for peak-capacity sessions ("consulting the oracle"), and lets ideas rest before prompting an LLM.
As frontier models become more token-efficient at lower effort levels, his weekly allocation now stretches from one day to three or four — but he isn't always better off: the forced waiting itself is part of the thinking process.
More from AGI Musings
- Robin Hanson's "Foom Liability": a robust policy case for AI liability incentives — NathanpmYoung · 2026-09-14
- Researchers map a roadmap toward genuine recursive self-improvement in AI — irinarish · 2026-09-14
- We're burning the planet to generate code engineers have to reject — joshuaamdamian · 2026-09-14
- Burkov: saying LLMs replace books is like saying students learn courses by quizzing professors — burkov · 2026-09-14
- Engineering veteran: AI safety debate lacks people who've shipped and failed at scale — billhilf · 2026-09-14
- tszzl pushes back on calling Elon Musk a decel: you've lost the plot — tszzl · 2026-09-14