EVERY CEO BUILDS AI CLONE FROM 30,000 EDITS

Shipper used 30,000 instances of his own editorial feedback to train an AI agent capable of mimicking his editing style and decision-making patterns. The model handles copyediting tasks—factchecking, grammar, consistency—traditionally performed by human editors. The company plans to double headcount while maintaining or reducing overall costs through selective automation. Rather than replacing editors entirely, Every views AI as handling repetitive editorial work, freeing staff for deeper reporting and analysis. Shipper highlighted a counterintuitive tension: writing with AI requires careful prompting and oversight that sometimes demands more effort than traditional writing. The process exposes what he calls a "dirty secret" in AI-assisted content creation—the technology isn't a simple productivity multiplier. The experiment raises questions about scaling editorial operations and the future of AI in newsrooms, though Every maintains human judgment remains central to final editorial decisions.