- LLMs have become reliable enough to run in automated feedback loops, leading to a 2x productivity boost but not a 10x improvement.
- The author uses LLMs for tasks with easily verifiable criteria, like generating code that meets explicit acceptance criteria.
- LLMs still struggle with subjective assessments like code maintainability and documentation quality, requiring heavy human iteration.
- The author instructs LLMs to avoid writing READMEs, docstrings, or comments to improve output quality.
- Further model improvements alone are unlikely to yield a 10x boost; productivity gains will come from better tooling and workflows.
- Vibe coding (generating code without fully understanding it) is explored for non-production tasks, but its long-term viability is uncertain.