An Anthropic engineering leader says Claude writes all the code for her 200-person team, a sign of how fast AI is changing the work of building software.
A top engineer at Anthropic, the maker of the Claude AI assistant, says her team no longer writes code by hand. Katelyn Lesse leads platform engineering at the company. She said all of her team's code is now written by Claude, an AI tool. In her words: "Literally all of it."
Lesse spoke on Tuesday at Grafana Labs' ObservabilityCON event. Her group handles Anthropic's core services, developer tools and product infrastructure. That makes it the engine room of the company. Her staff numbered 200 as of this summer. She said the era of human-written code is fully over for the team, both for building and for maintaining its systems.
People on the team have not left the process. They still design the systems and decide how to check that the work is right. Then they hand the actual coding to Claude.
Lesse gave an example. The company moved a major codebase, the body of code behind a product, from one programming language (Python) to another (Rust). Workers built a way to verify the final result. The rest of the building and monitoring was done by AI agents, which are programs that carry out tasks on their own. She called it "agents all the way down."
Anthropic already lets agents fix tests on their own, with no human involved. Letting them change the real codebase on their own is a bigger step. Lesse said engineering leaders are "terrified" and want to be careful, but she believes the line is starting to move, helped by better AI models over the past year.
Anthropic uses Claude Code across the whole company. In May, it said Claude wrote more than 80% of its production code. Last week it released a report saying it has hired for a smaller share of junior technical jobs as AI agents have done more work inside the company.
The company is also approaching an initial public offering, which is when a company first sells shares to the public. That could value it at nearly $2 trillion. Its revenue growth slowed in the third quarter of this year. Keeping that valuation may depend on persuading more companies to work the AI-first way Lesse describes.
The claim comes from one Anthropic leader speaking about one team, and Anthropic has a business interest in showing that its tools work. The material does not say how many bugs or problems the AI-written code has, or how much time and money it saves. It also does not say whether other companies can match these results.
Anthropic says AI writes all the code on one 200-person team, so expect software work and hiring to keep shifting toward AI.
A tech writer fed his Windows error logs to ChatGPT and learned his encrypted drive had no recovery key, a problem that could have locked him out for good.
Read more: ChatGPT read his laptop's error logs and found three hidden problemsA report says Anthropic could trade under the ticker ANT and cites $16.3 billion in revenue. Here is what that could mean for Claude users and investors.
Read more: Anthropic stock market debut: ticker ANT, $16.3B revenueA TikTok Shop affiliate says a strong first line, an honest demo and steady posting earned him six figures last year. Here is how it works and what it costs to try.
Read more: How a TikTok creator earns six figures with a 3-second opening line
