Beyond the hype, very few companies are taking agentic engineering seriously

The author argues that very few companies take agentic engineering seriously, instead falling into the trap of focusing on skills and subagents rather than solid fundamentals. Successful agentic coding relies on robust engineering practices from a decade ago, including comprehensive test suites, deterministic code rules, and senior code review. Consequently, testing and review become the actual pace-setting bottlenecks when AI writes the code. Open article →