Google engineer says Claude Code built in one hour what her team spent a year on (the-decoder.com)

In a striking demonstration of AI's rapid progress, Google Principal Engineer Jaana Dogan revealed that Anthropic's Claude Code was able to generate a distributed agent orchestration system in just one hour—something her team at Google had been laboriously developing for an entire year. While the output wasn't flawless and requires refinement, Dogan acknowledged its comparability to Google's prior efforts, marking a significant leap in AI-assisted coding capabilities. This incident underscores the swift evolution of AI tools, with predictions suggesting that by 2025, such systems could autonomously create and restructure entire codebases. Boris Cherny, the creator of Claude Code, emphasized the importance of enabling self-verification in coding tasks, which can significantly enhance output quality. He shared best practices for using Claude, such as initiating sessions in a planning mode and employing automation for repetitive tasks. This development not only showcases the remarkable capabilities of AI coding assistants but also raises questions about future collaboration between competing firms, as both Dogan and Cherny express optimism about the potential for collective advancement in the tech landscape.