Google engineer says Anthropic's AI matched team's year-long effort
Jaana Dogan, a Principal Engineer at Google, shared that Anthropic's Claude Code AI recreated what her team spent a whole year building—in just one hour.
"I'm not joking and this isn't funny," she posted on X, describing how the AI produced a prototype closely resembling their project after she gave it a simple summary.
What actually happened?
Dogan ran an experiment with Claude Code using only public info and a short description about coordinating multiple AI agents—a problem her team had wrestled with all year.
The AI quickly built a working prototype that followed their design patterns, though it still needed some tweaks.
How is Google reacting?
Even with these wild results, Dogan made it clear: at Google, Claude Code is only allowed for open-source work—not for internal work.
She also pointed out that tools like this aren't here to replace engineers but can help them move faster.
Why's everyone talking about it?
Dogan's post blew up with over four million views.
She later added that while the AI made a solid demo from their best ideas, real progress comes from humans and tech working together—not competing against each other.