Head of Claude Code: What happens after coding is solved | Boris Cherny

Boris Cherny describes a software workflow in which Claude Code writes all of his code, while he ships 10–30 pull requests a day and runs several agents in parallel. He says coding is more enjoyable because he no longer handles minutiae, and estimates that productivity per engineer has increased 200%. He expects coding to be nearly solved within a year or two, making software construction available to anyone rather than only trained engineers.

The next change is not simply faster code generation. Claude is beginning to identify ideas by reading feedback, bug reports, and telemetry, then proposing fixes and things to ship. Cherny predicts that everyone may become a product manager and builder as software-engineer boundaries blur. The episode introduces his work at Anthropic, the rapid growth of Claude Code, and the safety mission that drew him back after a short move to Cursor.

Claude Code’s growth reflects a broader shift from code generation to agentic work. Cherny describes a trajectory from coding, to tool use, to computer use, with safety as a central reason for developing capabilities in stages. Cowork extends that transition to nontechnical users by allowing an agent to act through tools such as Gmail and Slack. The team explored model boundaries through prototypes, research, and a deliberately minimal terminal interface rather than a heavy predesigned environment.