AIAIAI.FYI
LIVE

Anthropic says Claude handles about 90% of code writing and review on the Claude Code team

The New Stack reports the claim, and says the open question is how Anthropic verifies the work Claude produces.

What happened

Anthropic makes an AI assistant called Claude. The company says Claude now does most of the coding work for the team that builds Claude Code, a coding tool. That covers both writing the code and checking it.

Why it matters to you

If an AI can write and check most of a software team's code, companies may rethink how they staff and review software work.

The takeaway

Anthropic says its own team lets Claude do about nine tenths of its coding and reviewing.

What it means for you

If you build software

Teams using AI to both write and review code should look at what independent checks sit behind that review.

If you run a business

The claim is a vendor's report about its own team, so it is not proof of the same results elsewhere.

If you invest

Anthropic reports the 90% figure itself; no financial data accompanies it.

Get stories like this every morning