‘I am not joking,’ says Google engineer as Anthropic’s Claude Code recreates a year of work in just one hour

Jaana Dogan, Google principal engineer on the Gemini API team, shared her experience testing Claude Code, an AI tool created by Anthropic.

'I am not joking,' says Google engineer as Anthropic’s Claude Code recreates a year of work in just one hour
'I am not joking,' says Google engineer as Anthropic’s Claude Code recreates a year of work in just one hour

A recent post by a senior Google engineer has ignited strong reactions across the tech industry, raising fresh questions about how fast AI tools are changing software development. The engineer revealed that an AI coding tool built a working system in one hour something her team had spent nearly a year developing. The post quickly went viral online, catching the attention of developers, managers, and tech leaders around the world.

One Year of Work done in a Hour

Jaana Dogan, a principal engineer at Google working on the Gemini API, shared her experience testing Claude Code, an AI tool developed by Anthropic. Her team had been building a complex system to manage multiple AI agents which is a task that involved months of planning, discussions, and architectural decisions.

Out of curiosity, Jaana Dogan described the same problem to Claude Code using a short, high-level prompt. To her surprise, the tool produced a working prototype in about an hour. The structure and approach were very similar to what her team had created after months of effort.

She made it clear that the AI’s output was not ready for real-world use. Still, the speed and accuracy of the result impressed her.

Important Clarifications

Jaana Dogan was careful to explain that the AI did not magically replace her team’s work. The prototype still needed refinement, testing, and human judgment. She also noted that Google limits the use of third-party AI tools like Claude Code to open-source projects, not internal systems.

Another key points is that the AI worked well because the problem was already well understood. The clearer the problem description, the better the AI’s results. This means experience and expertise are still essential.

What This Means for Developers?

The incident has shifted the conversation in tech circles. Many developers now believe that AI can dramatically speed up early stages of development especially prototyping and system design.

However, experts warn that AI-generated code still needs careful review. Issues like security, performance, and long-term maintenance cannot be left to automation alone.

Humans Still Matter Most!

This post highlights how AI is becoming a powerful helper rather than a replacement. Tools like Claude Code can save time and reduce repetitive work, but human engineers remain responsible for decisions, quality control, and long-term vision.

Thereofore for now, the future of software development looks less like humans versus AI and more like humans working faster with smarter tools.

This article was first uploaded on January five, twenty twenty-six, at eighteen minutes past four in the afternoon.