[Claude Helps Anthropic Develop Next-Generation Claude, Surges from Less Than 1% to 26% in Six Months] Beating AI Newsflash: Anthropic recently disclosed that Claude is now leading 26% of AI development tasks, up from less than 1% in February. Over 90% of development tasks are now completed with at least some collaboration between humans and Claude. The term 'leading' here has a clear definition: humans only need to tell Claude the ultimate problem to solve, and it can handle most of the process independently, with humans reviewing the final results. Currently, no type of development task can be fully entrusted to Claude. This figure goes beyond Anthropic's previously announced '80% of production code written by Claude.' Code is just one part of development; this statistic also includes model training, experiments, evaluations, infrastructure, and the entire development workflow. On Anthropic's most commonly used internal platform, approximately 30,000 agents are conducting research and engineering simultaneously at any given moment. [Original Link]