Svmuu News: Anthropic’s latest report on large-scale agent programming shows that, with the assistance of Claude Code, the importance of programming skills themselves has declined, while management and domain knowledge have become key factors influencing outcomes. Claude Code users average 20 hours of weekly active time. Across the top ten major professions, the success rate of non-programmers using AI agents to write code has narrowed to within 7% of that of professional software engineers, with managers achieving a success rate slightly higher than that of professional programmers. Users drive approximately 70% of planning decisions, while AI agents handle about 80% of specific execution decisions. A single expert-level session can trigger 12 actions and generate 3,200 words of output—more than five times that of a beginner-level session. The abandonment rate for beginner sessions after encountering an error reaches 19%, while intermediate and expert users have an abandonment rate of 5% to 7% when encountering issues. Over a 7-month period, the proportion of debugging sessions used to fix broken code dropped from 33% to 19%, while the proportions of sessions dedicated to direct deployment and execution, data analysis, and writing non-code documentation doubled, and the average estimated value of tasks increased by approximately 25%.