Artificial Intelligence
Claude Code’s Creator Says the Engineer is Becoming a Builder
Boris Cherny built Claude Code, and he hasn’t written a line of code in over six months.
That’s not a confession. In a with Casey Newton, the creator of Anthropic’s coding agent said the title “software engineer” could start disappearing by the end of this year, dissolving into something he keeps calling a “builder.” Coding, at least the kind he does, is “effectively solved.” He isn’t mourning it. He’s building all day; the model does the typing now.
The headline writes itself as a jobs-apocalypse story, and most coverage took it there. That’s the wrong read for anyone actually running an operation. Cherny didn’t describe the end of work. He described the rise of , one defined not by what they can code but by what they can build with an agent sitting at the center of everything they do.
The tell isn’t who’s losing the job — it’s who’s winning the hackathon
Buried in the interview is the detail that matters more than any prediction about headcount. Anthropic ran a hackathon for the Opus 4.7 release, and the winners “were largely not professional engineers.” An electrician. A doctor. A carpenter who used Claude to build an app. Cherny saw the same thing at the 4.6 hackathon, and his read was blunt: “the people who get the most value out of Claude Code are not the people I’d expect.”
That should land hard if you’ve been waiting until you’re technical enough to start. The barrier was never the code. The carpenter didn’t beat the engineers because he learned to program over a weekend. He beat them because he knew exactly what he wanted to build and pointed an agent at it.
You can watch the same thing happen inside Cherny’s own team. His manager, Fiona, hadn’t written code in 15 years. She codes now. Kat, the product manager, codes. Megan, the designer, codes. “Everyone on the team codes,” he said. “You don’t have to be an engineer anymore.” The role isn’t shrinking. It’s absorbing everyone who used to stand outside it.
This is the part the layoff framing misses. When coding stops being the bottleneck, the scarce skill becomes judgment about what to build and the discipline to run the system that builds it. That skill has nothing to do with a CS degree. It’s the operator’s skill, and it’s been hiding in plain sight inside accountants, marketers, and tradespeople this whole time.
Put Claude at the center, or leave it in the corner
The sharpest thing Cherny said wasn’t about coding at all. It was a history lesson. He pointed to a 1990 on companies adopting the personal computer. The ones that got more productive threw out the filing cabinets and put the computer at the center of how they worked. The ones that didn’t kept a computer in the corner, used it for one thing, and saw no gains.
Then he described Anthropic in exactly those terms. “We organize everything around Claude.” Questions about the codebase: ask Claude. Filing an expense: ask Claude. The next company holiday: ask Claude. “The companies that are really getting it put Claude right at the center, not on the outskirts somewhere.”
That’s the operator thesis, stated by the person who built the tool. The leverage doesn’t come from the model. Everyone has the same model. It comes from rebuilding your operation around it instead of bolting it onto the side. A chat window open in another tab is the computer in the corner. A workspace where the agent runs your research, your drafting, your scheduling, your client delivery is the filing cabinets in the dumpster.
Cherny’s own habit makes the point. He still personally answers user complaints on X and Threads, and he automated the collection of that feedback with a loop in Claude Code that he “moved to a routine that runs every 30 minutes.” That’s not how an engineer thinks about a side task. That’s how an operator thinks about a system: the boring, recurring work runs itself so the human time goes to the part that actually needs a human.
The window is open and it’s the operator’s to walk through
Cherny’s advice to a new CS graduate wasn’t “learn to prompt.” It was “go start a startup. There has never been a better time in history to do it.” His reasoning: “You and your agents can build a giant company.” He’s watching customers run real businesses with one, two, three people, the same way Claude Code itself started with a handful.
He won’t predict whether the world ends up with more engineers or fewer. But on builders he’s certain: “I think there will be 100 times more of them than there are today.”
The number is a guess. The direction isn’t. The work is shifting from the people who can write the code to the people who can decide what’s worth building and stand up the system to build it. That person doesn’t need permission, a title, or a degree. They need a real operation with the agent at the center, and they need to start before the title catches up to what they’re already doing.
The carpenter didn’t wait. Neither should you.
