Anthropic published new internal performance dashboards showing that Claude now contributes to 26% of the work required to build its next iteration, compared to under 1% earlier this year. Over 30,000 AI agents run simultaneously inside the company, with safety monitors blocking roughly 1 action per 47,000. The update arrives shortly after industry discussions about slowing AI development, highlighting how quickly frontier models are evolving.
About the Author
