How much of a newsletter editor’s job can Claude do?

Feature

Platformer’s experiment is deliberately personal: after earlier self-automation attempts, a staff writer spun up a Claude Fable 5 agent to imitate editor Casey Newton. The bot’s name — Claudeasey Newton — is half joke, half product test for whether a frontier model can stand in for a human edit pass on a tech newsletter.

The author’s scorecard is blunt: real Casey comments land close to 95% helpful, while the bot’s comments were closer to 70% off the mark — leaving roughly 30% useful enough that the speed still felt worth it. Platformer is winding the Claudeasey experiment down; Casey stays boss. The writer still expects to use LLMs for some editing tasks previously handed to a human — while remaining relieved that Claude is, for now, only a medium-quality editor.

“the human-intensive, provenance-rich, sometimes artisanal part of the economy where the human aspect is part of the value of the good or service itself.”