unknown

A software engineer shares their workflow for using LLMs (primarily ChatGPT Codex) to build production software, detailing how they built a React data library called fate with 80% AI assistance. The author uses multiple concurrent AI sessions generating 4 solutions per problem, heavily edits AI-generated code, and maintains strict quality standards. They discuss the spectrum between never using AI and relying on it completely, predict that 10x productivity through AI orchestration will become the baseline expectation, and reflect on how LLMs are transforming software frameworks, personal computing, entertainment, and knowledge sharing.