Encoding Team Standards

This article by Rahul Garg, a Principal Engineer at Thoughtworks, argues that AI coding assistants produce variable results because team standards are tacit knowledge held by individuals. He proposes encoding these standards—for generation, refactoring, security checks, and code review—into versioned, reviewed, and shared artifacts. This transforms tribal knowledge into executable instructions, ensuring consistent quality regardless of who prompts the AI. The article discusses the consistency problem, executable governance, surfacing tacit knowledge, and treating standards as shared infrastructure for AI-assisted development.