How a Cloud Architect Used Classic Copywriting Rules to Fix AI Writing

A cloud architect and AI writing collaborator found that using AI in its default mode produced generic, detectable prose that readers increasingly skip. Drawing on five renowned copywriters — including Ogilvy, Zinsser, and Sugarman — he embedded specific craft rules directly into his AI agent's instructions. Techniques like cutting throat-clearing sentences, using concrete specifics instead of vague summaries, and leading with loss framing significantly improved output quality. Research from ICLR 2024 supports the approach, showing that RLHF-trained models produce less diverse text than their base versions, explaining AI writing's tendency toward a single bland register. The author notes the method raised his baseline quality and cut editing rounds from ten to two or three, though human oversight of the final draft remains essential.