Top engineers at Anthropic and OpenAI now rely entirely on AI to write code, while an open source AGI leader says most software this decade will be machine-generated, reshaping how projects are built and maintained.
AI-generated code is moving from assistance to dominance, and open-source leaders say the shift will redefine how software is built, maintained and contributed to.
At Anthropic, engineers have largely stopped coding by hand. Boris Cherny, Head of Claude Code, says 100% of his work is now AI-written, while company-wide output is estimated at 70–90%. Claude Code even writes about 90% of its own codebase.
“For me personally, it has been 100% for two+ months now, I don’t even make small edits by hand. I shipped 22 PRs (pull requests) yesterday and 27 the day before, each one 100% written by Claude,” Cherny wrote.
CEO Dario Amodei warned the industry may be “six to twelve months away from AI handling most or all of software engineering work from start to finish”.
OpenAI shows similar behaviour. Researcher Roon said: “100%, I don’t write code anymore… Programming always sucked… and I’m glad it’s over.”
The open-source ecosystem is preparing for structural change. Himanshu Tyagi, co-founder of open-source AGI company Sentient, said: “A large part of the code that will be shipped over the next 10 years will be written by AI… Claude Code is the breakthrough product that has made that happen.”
Adoption remains lower elsewhere — Microsoft reports ~30% AI-written code and studies show ~29% on GitHub — but rising.
Engineers report faster releases and less drudgery, even as hiring shifts towards generalists and models still make “subtle conceptual errors”.
For open source, AI agents could accelerate contributions, automate maintenance and lower barriers — while raising fresh questions around ownership, licensing and accountability.
