Google Engineer: Claude Code Replicates a Year of Team Effort in One Hour

21:24, 03 January

Edited by: Veronika Radoslavskaya

On January 2, 2026, Jaana Dogan, a Principal Engineer at Google, shared an account on X regarding her experience with Anthropic’s Claude Code. She stated that the AI agent drafted a distributed agent orchestration system in roughly one hour—a task her team had been addressing for a year.

Key Facts and Details

The Scenario: Dogan provided the tool with a three-paragraph prompt outlining a specific problem. She noted that while her team had spent a year discussing the architecture and trying to reach a consensus on this project, the AI provided a comprehensive draft and implementation in a single session.
Architecture and Code: Dogan characterized the resulting architecture and code as highly impressive, despite not providing the AI with any internal Google data. She clarified that the output was not perfect and required further refinement, but it effectively solved the architectural challenge that had previously stalled her team.
Internal Policies: At Google, engineers are permitted to use Claude Code for open-source contributions, but company policy strictly prohibits its use on internal proprietary systems.
Response Regarding Gemini: When asked about Google’s progress in this area, Dogan commented: “We are working hard right now. The models and the harness.” She noted that the AI industry is not a zero-sum game and that rapid innovation across the field serves as a motivator for her team.