Boris Cherny is 35, holds the title head of applications at Anthropic, and is the person who created Claude Code — the coding agent that can write code, run commands and work on files on a user's behalf.
He joined Anthropic in 2024 after nearly seven years at Meta, where he worked across Instagram, Facebook Groups and the code quality team.
The prototype
He spent his first month at the company poking at the models to see what they could do. The turning point was a small command-line tool that let Claude reach his computer's operating system directly.
His test was deliberately trivial: figure out what music he was listening to.
The model wrote code to answer the question without being told to write any. "The model was given this tool, and it figured out how to use it to answer this question that I had that I wasn't even sure it could answer," he said on Lenny's Podcast.
That side project became Claude Code, which passed $2.5bn in run-rate revenue in February.

The number that says more than the revenue
Anthropic later built Claude Cowork, an agent for general office work that manages files and automates tasks. It built it in ten days, using Claude Code.
Ten days is the claim worth sitting with. It is not a statement about how good the output is — it is a statement about how short the distance has become between deciding to build something and having it exist.
Cowork exists because Cherny noticed people using a coding tool for work that was not coding. He uses it himself for project management and for paying parking tickets.
What that looks like outside Silicon Valley
At the Mid-Ohio Food Collective, non-technical staff use Claude Code routinely. Its director of nutrition built a game — described as something like Tinder for food — to help families pick healthier options based on preference.
The idea was not new. Mark Mollenkopf, the organisation's director of software engineering, still has photographs of a whiteboard sketching the concept in 2017. For nine years it lost to higher priorities.
"We're not going to put it in production today yet," Mollenkopf said, "but it's that unlock for us to be able to democratize great ideas that we've always had."
That is the quiet shift. The constraint was never the idea. It was that someone had to be free for three weeks to build it.
The man who built it is not selling a comfortable story
Asked about the effect on jobs, Cherny did not reach for reassurance. "I think for the job market, in particular for engineers, it will be disruptive," he told CNN earlier this year. "I hope that over the long term, it's amazing for humanity and what it unlocks."
Note the two halves: disruptive is stated, the good outcome is hoped for. He also describes his own relationship with the craft in terms that make the first half personal. "When I write code, I have an emotional relationship with it… I dream about it."
Why any of this matters commercially
Ara Kharazian, a senior economist at the spend-tracking firm Ramp, dated the acceleration in business AI spending to late 2025, after Claude Code's general release in May. It was, he said, "the first commercially advanced version of AI that was directly productivity enhancing for a specific segment of jobs".
Anthropic's expected initial public offering — reported at more than a trillion dollars — rests on whether Claude becomes indispensable to office work rather than just to engineers.
Willy Lee of the venture firm Neostellar framed how investors read it: "When people think about the IPO, they're thinking, 'Okay, how much room left does Claude Code have to grow?'"
The risks are real: competition from OpenAI, dependence on a small number of large customers, and a valuation that leaves no room for a slow year. We have covered the IPO pricing separately. This is the product underneath it.




