A principal engineer at Google has revealed that Anthropic's Claude Code, an advanced AI-powered coding tool, generated a complex distributed agent orchestration system in just one hour, a task her team at Google had been grappling with for over a year.
Jaana Dogan, who works on the Gemini API, shared her experience on social media platform X, stating: “I'm not joking and this isn't funny. We have been trying to build distributed agent orchestrators at Google since last year. There are various options, not everyone is aligned… I gave Claude Code a description of the problem, it generated what we built last year in an hour.”
Rapid prototyping highlights AI's accelerating role in software development
The system in question coordinates multiple AI agents, a challenging area involving intricate design choices and trade-offs. Dogan clarified that her prompt was brief, merely three paragraphs, and based on publicly available ideas, as she could not share proprietary Google details. The output, while not production-ready and requiring further refinement, served as a strong starting point and impressed her with its design recommendations.
She emphasised that the underlying concepts are not novel, and her teams had already developed several versions with differing compromises. However, AI tools like Claude Code enable swift prototyping from scratch, free from legacy constraints often present in large organisations.
Industry not zero-sum, says Dogan amid praise for competitor
Dogan praised the competitor's work, noting, “This industry has never been a zero-sum game... Claude Code is impressive work, I’m excited and more motivated to push us all forward.” She added that Google permits use of such tools only for open-source projects, with internal efforts focused on Gemini models and infrastructure.
