Anthropic’s IPO filing shows how AI is changing the job of a software engineer

Anthropic's own AI now writes most of the software inside the company. Claude produced more than 90% of the contributions to Anthropic's codebase in September, according to the company's S-1 filing with the US Securities and Exchange Commission.

The 335-page document, filed on October 26, sets Anthropic up for a Nasdaq listing under the ticker "ANT", possibly as early as next month.The filing also shows how the job of an Anthropic engineer has changed. Its software engineers now mostly direct and review Claude's output instead of writing code line by line. The numbers back it up: average daily code contributions per engineer in the July to September quarter were more than 45 times their average daily level in 2024.

Anthropic is using Claude to build the next version of Claude

Coding is only one part of a wider shift. Anthropic says it now relies on its models to research, design and develop the models that come after them, a process the company calls closing the loop. Claude runs experiments, generates training data, and automates evaluations and safety audits. It has also helped with algorithmic improvements that let Anthropic get more out of its computing resources."Claude itself has become one of the most powerful drivers of our progress," the company wrote in the filing.

That reliance shapes hiring too. Anthropic's workforce grew from about 940 employees at the end of 2024 to roughly 5,300 by late September, and the company says its strategy is to keep growing the business without adding staff at the same pace. A recent internal report noted that it has been filling a smaller share of junior technical positions as AI agents take on coding, bug fixes and similar tasks. Regrettable attrition, meanwhile, stood at about 1% for the 12 months ending September 30.

Anthropic CEO predicted the 90% mark more than a year and a half ago

The figure will not surprise anyone who has followed CEO Dario Amodei. In March 2025, he said AI would be writing 90% of code within three to six months, and could be writing essentially all of it within a year. By October that year, he told an audience at Salesforce's Dreamforce conference that the number already held true within Anthropic. He also argued that companies might end up needing more engineers, since each one becomes far more productive with AI doing the bulk of the typing.Some teams inside the company have gone even further. Boris Cherny, who heads Claude Code, said in June that he had not written a line of code by hand in about eight months.The filing pairs all this with a note of caution. Among its long list of risk factors, Anthropic warns that its AI systems may behave in unintended, misaligned or autonomous ways that could damage the business. That is the same system now responsible for nine in every ten contributions to the company's codebase, and increasingly, for helping build the next Claude.