Anthropic: A Study of 400,000 Sessions Shows Non-Programmers Using Claude Code Achieve Success Rates Close to Software Engineers

Odaily reported that Anthropic's latest large-scale intelligent agent programming report shows that with the assistance of Claude Code, the importance of programming skills themselves has diminished, while management and domain knowledge have become key factors influencing outcomes. Claude Code users are active for an average of 20 hours per week. Among the ten major occupations, the success rate gap between non-programmers using intelligent agents to write code and professional software engineers has narrowed to within 7%, with management roles having a slightly higher success rate than professional programmers. Users lead approximately 70% of planning direction decisions, while the intelligent agent executes about 80% of specific implementation decisions. A single expert-level session can trigger 12 actions and output 3,200 words, more than five times that of a novice-level session. Novice sessions have a 19% abandonment rate after encountering errors, while intermediate and expert users have abandonment rates between 5% and 7% when facing problems. Over seven months, the proportion of debugging sessions for fixing broken code dropped from 33% to 19%, while the proportions for direct deployment and operation, data analysis, and writing non-code documentation doubled, and the average estimated value of tasks increased by approximately 25%.