AI coding agents are following the same historical pattern as compilers and high-level languages — abstracting away lower-level concerns so developers can focus on higher-level problems. The emerging practice of 'agentic engineering' involves defining specs, providing context, and enforcing tests rather than writing code directly. The author argues that rather than rejecting this shift, developers should embrace and steer it — especially by building vendor-neutral, open-source/locally-hosted alternatives to Big AI tools — to retain agency over their craft and avoid ceding control to companies with ill intent toward working coders.
Sort: