Anthropic says its engineers write 250% more code with its internal Claude Code tool

281459

“Nice idea except that when your product is a model and you have infinite free tokens inside as part of the cost of doing business then you don't need to "over-control AI usage" but for the rest of the world it's not the same.”

Dan Jeffries@Dan_Jeffries1TECH#1711

Story Brief

Platformer reported that Anthropic's Claude Code leader, Boris Cherny, said code volume per engineer rose after the company adopted Claude Code. On X, a reply from Dan Jeffries argued that Anthropic's internal economics are different from those of outside teams that have to pay for usage, while Business Insider and an Anthropic engineering postmortem add context on both rapid internal adoption and the company's own recent quality issues.

“@Dan_Jeffries1 yep, all of us are so concerned about cost now. 😟”

Rohanpaul Ai

Source Posts on X

Cluster sources

Daniel Jeffries@Dan_Jeffries1TECH

Nice idea except that when your product is a model and you have infinite free tokens inside as part of the cost of doing business then you don't need to "over-control AI usage" but for the rest of the world it's not the same. A team spending for those tokens on the outside burns their budget in three months instead of a year, so not realistic in the least unless you are running fully open source local models.

Anthropic Reports 250% Jump In Code Per Engineer With Claude Code

Related Stories

Source Posts on X

Useful Links

Digg Deeper