By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
WorkOSBoris Cherny: Claude Code & the Future of Engineering | Acquired Unplugged presented by WorkOS
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
At a glance
WHAT IT’S REALLY ABOUT
Boris Cherny on Claude Code, productivity, and builders replacing roles
- Boris Cherny said Claude Code began as an accidental-feeling prototype within Anthropic’s Labs team to build a frontier coding agent and learn how to evolve the model alongside a real product.
- He explained that Anthropic’s emphasis on coding and tool/computer use is linked to its AI-safety mission, because studying the model “in the wild” requires useful real-world interactions and coding is a clear interface for that.
- He attributed the major jumps in Claude Code capability primarily to model improvements (citing Sonnet 4/Opus 4 and later Opus 4.5), with additional work improving the experience across CLI, desktop, mobile, and integrations like Slack and GitHub.
- He described large internal productivity shifts at Anthropic, including a previously shared claim of about 3X more code per engineer and onboarding time dropping from weeks to roughly two days by relying on Claude for guidance and execution.
- He argued organizations should evolve toward generalist “builders,” fund fewer people per project but provide abundant tokens, and expect “taste” advantages to erode as models get better—leaving values as the hardest remaining human input.
IDEAS WORTH REMEMBERING
5 ideasClaude Code emerged from a Labs prototype to close a perceived product gap for capable models.
He said he joined a prototyping-oriented Labs team to find “the next big product” while simultaneously pushing model capabilities, and that a “product overhang” made a full coding agent feel like a frontier product worth attempting.
Coding was prioritized because it supports Anthropic’s safety mission and real-world study of model behavior.
He framed coding as a way to study models “in the wild,” tying it to Anthropic’s stated core purpose of AI safety, and argued coding is a natural interface for how software models interact with the world.
The biggest driver of Claude Code’s effectiveness was underlying model improvement, not just product polish.
While he noted improvements in the harness and multiple form factors, he attributed the major step-changes in usefulness—how much code it could write—to new model releases (citing Sonnet 4/Opus 4 and later Opus 4.5).
Claude Code materially changed internal productivity and onboarding dynamics at Anthropic.
He claimed internal output increased sharply, citing a previously shared figure of roughly 3X code per engineer (while saying it’s now outdated and higher), and described ramp-up for new hires dropping to about two days because people can ask Claude instead of other engineers.
He expects engineering work to shift from writing code to orchestrating automated loops that drive the model.
He described his own progression from IDE autocomplete, to prompting multiple concurrent Claudes, to building “loops” that prompt Claude automatically—positioning loop-writing as the next abstraction layer in engineering work.
WORDS WORTH SAVING
5 quotesLike, if, if you ask, like, a random person at Anthropic, you just, like, pull them over in the hallway and ask, "Hey, why are you here?" They're gonna say, "AI safety."
— Boris Cherny
In November, I uninstalled my IDE 'cause I wasn't using it.
— Boris Cherny
Now it's actually leveled up, I think, again to the next level of abstraction where I don't prompt Claude anymore. I have loops that are running. They're the ones that are prompting Claude and kinda figuring out what to do. My job is to write loops.
— Boris Cherny
I think, like, we're, we're kinda starting to see all the roles merge and kind of melt into one builder.
— Boris Cherny
I th- I think the final thing that we're gonna be teaching the model is values, and s- how to be, you know... The, the way that we teach our kids how to be good people, we're gonna teach the model how to be a good model.
— Boris Cherny
SPEAKERS
Ben Gilbert
hostCo-host of Acquired; tech/business interviewer and former Microsoft employee.
Boris Cherny
guestMember of Technical Staff at Anthropic and builder of Claude Code, with prior experience in developer tools at Meta.
David Rosenthal
hostCo-host of Acquired; interviewer focused on company/product strategy and engineering.
High quality AI-generated summary created from speaker-labeled transcript.
