Linear data: 53% of merged code now ships with AI review only, no second engineer
rseroter · x · 2026-11-20
Les Ottolenghi's essay examines where human review still matters as AI writes more code.
Linear's data: In October, 53% of merged code changes shipped after only an AI review plus the author's sign-off, versus 1% in January. Merged changes rose 153% while human authors grew just 22%. Review requests per author rose 38% in six months, and median change size doubled from 47 to 108 lines.
Microsoft/CMU study (Nov 13): surveying 2,590 developers, 71% said review has shifted toward design and intent; AI review doesn't clear the queue on its own.
Secure Code Warrior AI Trust Index (Nov 17): 30+ models produced 0.24–3.66 vulnerabilities per 1,000 lines — a 15-fold spread.
The core test: ask your engineering leader which changes can never merge without a second person and whether the pipeline enforces it. If it's a habit rather than a rule, your review process was built for a world where humans wrote the code.
More from coding & agent
- Raycast launches MCP: let Claude, Codex and any agent drive your Mac launcher — dean_rie · 2026-11-20
- A 6-model board for decisions: hide opinions, let Claude chair and adopt the sharpest argument — aryanXmahajan · 2026-11-20
- Zed Delta agents can now run background terminals and get notified on completion — dean_rie · 2026-11-20
- Gemini app for Mac ships computer use, automating tasks across folders, apps and the web — GeminiApp · 2026-11-20
- Claude builds its own tooling mid-task, cutting overnight jobs to one hour — DaniyarQQQ · 2026-11-20
- llms.txt "never really existed": Mintlify CEO on agents now driving 65% of doc traffic — zeeg · 2026-11-20