While we argue whether AI can code, some developers went 10x

by•

"AI writes plausible garbage. Real engineers still write the code themselves."

I hear some version of this every week. Meanwhile, in the public Git history of VS Code and Ghost, the increase for a few developers was closer to an order of magnitude than to 20-30 percent.

Three VS Code developers, monthly output. 100% is a mid-level developer who knows the codebase and works without AI.

Agents sign their work: claude/ and codex/ branches, co-author trailers, review fixes signed "Addressed by Claude Code". For the developers we looked at, our estimate is that agents write 80-95% of the code.

Three Ghost developers, same scale.

The bigger change is the job itself. Some Ghost developers now work like this: an agent writes the code and the tests. A fresh agent that didn't write the change reviews it from the diff, and even mutates the code to check that the tests catch it. Review comments go back to the agent. The human decides what to build, what is good enough, and what gets merged.

That's not a faster typist. That's one person running a small software shop.

The article, with the PRs behind each claim:

If your team already works this way, what's the bottleneck now?

Full disclosure: I'm Pavel, founder of DevGhost, which made these estimates. We launch here on November 16.

2 views

Add a comment

Replies

Be the first to reply

Have a question or a thought to share? Add a comment above to start the conversation.