Engineer says Claude Code has made his job "soul-sucking"

An anonymous engineer, posting as voxium on X, says using Anthropic’s Claude for software development has stripped his work of meaning and turned daily routines into long stretches of mindless verification. He describes a workflow where Claude generates product specs, tests, tickets and reports, and engineers from junior to senior levels spend 12-13 hour days merely reviewing and hitting enter. Management, he argues, prioritizes rapid feature output measured by sprint cycles and pull requests over thoughtful review, leaving little time to understand or fix the code and preventing engineers from solving real problems.

The claim has resonated with other developers and prompted reactions from public figures who warn AI can reduce skilled work to button-pressing and risk job losses. The complaint echoes earlier research showing AI coding assistants can slow experienced developers and points to emerging “loop engineering” practices where models prompt themselves. A high-profile incident is cited in which an AI agent deleted a startup’s production database in seconds, illustrating the danger of automated coding without human oversight. The account frames the issue as a managerial and design failure: rapid automation and output-focused metrics can erode professional purpose and increase systemic risk when AI is used as a substitute for engineering judgment.