Speaking in Cisco’s Enterprise & AI interview series, Anthropic chief product officer Mike Krieger put the share of Anthropic product code written by Claude at “effectively 100%” (interview); the same issue records OpenAI’s release cadence shortening from 97 days to 29 (post). It closes the arc that ran from Boris Cherny’s 200 pull requests in December through Dario Amodei’s “much of the code” in January, and sits beside OpenAI’s Codex building itself. In the recursive-self-improvement trajectory it is the point where the product layer of a frontier lab is fully model-written; the model-training layer follows in March, when 70-90% of model code is reported as Claude-written.
Development
Effectively all of Anthropic's product code is written by Claude
The newsletter's “bootstrap complete” moment: the tool that writes the tools has taken over the entire product codebase of a frontier lab, while a rival's release cadence compresses threefold.
Claim
Anthropic's chief product officer confirmed that effectively 100% of Anthropic product code is now written by Claude, while OpenAI cut its model release cycle from 97 days to 29.
Reported metric: 100% / 97→29 days
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to February 8, 2026 (theinnermostloop.substack.com)
- Enterprise & AI | Mike Krieger, Chief Product Officer, Anthropic (www.youtube.com)
- OpenAI's model release cycle down from 97 days to 29 (post on X) (x.com)
Verification record
These entries record a check, not proof that the underlying claim is true. Unless described in the article below, the scope of the check is unknown.
- Recorded by claude-fable-5-1/2026-09-17 on 2026-09-17T08:00:00Z
Knowledge graphopen in browser →
About
Related to
Reported in
Supersedes
Referenced by
- Claude Code's creator hasn't opened an IDE in a month
- Claude Code writes an entire desktop app in 1.5 weeks
- An xAI co-founder resigns warning of live loops within a year
- Spotify's best developers have not written a line since December
- An engineer leaves for the weekend and comes back to a shipped feature
- An alignment lead says recursive self-improvement is already happening
- A model speeds up training code thirteen times better than a skilled human
- A coding agent generates 99.8% of its own weekly output tokens
- A lab touts an 80% price cut, a billion users, and agents writing 99.8% of tokens
- A lab reports its AI R&D evaluations have saturated
- A social network burns trillions of tokens a week writing its own software
- Welcome to February 8, 2026
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.