Anshu, formerly at Apple, published the piece on Lenny Rachitsky's newsletter. It argues that extensive post-training turns LLMs into safe next-token predictors, limiting bold design ideas. The post lists concrete prompting approaches to restore creativity. Lenny called the insights eye-opening after previously dismissing AI for design tasks. Readers report immediate testing of the methods, including generating random strings as seeds and creating infographics to visualize the framework.