‘I Don’t Write Code Anymore’: Anthropic, OpenAI Engineers Say AI Now Writes 100% of Their Code

Software engineers are used to shipping code. Now, some of the world’s top AI engineers say they don’t write it at all. Instead, AI does.

At Anthropic and OpenAI, engineers say AI now writes 100% of their production code, a shift that signals a big change in how software is built and what the future of programming jobs may look like.

The revelation gained momentum when Boris Cherny, the head of Claude Code at Anthropic, admitted on X that he hadn’t written any code by hand in over two months. Responding to AI researcher Andrej Karpathy, Cherny detailed a level of productivity that would have been physically impossible for a human just a year ago.

“For me personally it has been 100% for two+ months now, I don’t even make small edits by hand,” Cherny wrote in a post on X. “I shipped 22 PRs yesterday and 27 the day before, each one 100% written by Claude.”

Across Anthropic, the change is nearly universal. Cherny noted that “pretty much 100%” of the company’s code is now AI-generated. This isn’t just about simple scripts; Cherny’s team reportedly built Cowork, a new file-management agent for non-coders, in just about a week and a half by using Claude Code to write the software.

OpenAI joins the ‘zero-code’ club

The sentiment is echoed across the street at OpenAI. Roon, a prominent (though pseudonymous) researcher at the lab, confirmed that the manual era of programming has reached its finish line. 

In a reflection post on X, he said, “programming always sucked. it was a requisite pain for ~everyone who wanted to manipulate computers into doing useful things and im glad it’s over. it’s amazing how quickly I’ve moved on and don’t miss even slightly. im resentful that computers didn’t always work this way.”

When asked what percentage of his work is now handled by AI, the answer was absolute. “100%, I don’t write code anymore,” Roon posted on X. 

A 6-month countdown for the rest of us?

While Anthropic and OpenAI are the “Patient Zero” of this trend, Anthropic CEO Dario Amodei believes the rest of the world is next. 

Speaking at the World Economic Forum last month, Amodei predicted that the entire software industry is just “six to twelve months away from AI handling most or all of software engineering work from start to finish,” Fortune reported.

However, the transition isn’t without “slop.” Engineers admit the AI still makes mistakes, such as leaving behind “dead code” or overcomplicating simple logic. To combat this, Anthropic engineers now use the AI to review itself.

“At Anthropic we use claude -p for this on every PR and it catches and fixes many issues,” Cherny explained on X.

The new ‘generalist’ workforce

The idea of AI writing most code has immediate implications for the software job market, especially for entry-level engineers. Fortune noted that automation raises questions about the future of junior roles, which have traditionally been the training ground for new developers. 

The shift is also changing hiring priorities inside Anthropic itself, with more emphasis on adaptable generalists rather than narrow specialists.

Cherny says his team has shifted away from seeking specialists who master specific programming languages. Instead, they want generalists who can direct the AI’s creative and architectural decisions.

“Not all of the things people learned in the past translate to coding with LLMs,” Cherny wrote. “The model can fill in the details.”

A survey shows the AI workplace training gap: 89% of employees use AI at work, but only about 33% receive employer-provided training.