Developer proposes a 12-level AI coding autonomy scale based on who does the work
nimareq · reddit · 2026-11-05
A developer proposes a 12+1 level "AI development autonomy scale" for hiring clarity and team alignment, organized around "who does the work" rather than "what can AI do": Level 0 human-led; 1-4 (AI in the loop) inline completion through guided multi-file edits; 5-8 (human in the loop) supervised agents, delegated issues, feature delegation, and parallel agents; 9+ (human in the void) AI handling authorized routine changes. Each level maps AI/human roles, loop duration, context (local/cloud), and interface (IDE/CLI/Chat), with optional gamified metrics like LOC and review frequency.
More from coding & agent
- Engineer says manual tweaks after Claude optimization made EC2 CPU 50x faster than M4 Max — _Stocko_ · 2026-11-05
- A Rust todo app taking 14GB? Compile caches, not the app, eat ~13GB — lxfater · 2026-11-05
- Remote MCP: managing your devices remotely via the MCP protocol — Latter-Storage9610 · 2026-11-05
- MCP over HTTP: validating OAuth tokens isn't enough—check the audience claim — alvelda · 2026-11-05
- Short RL phase on GLM 5.3 Flash costs $372 vs $1,894 for notes, breaks even in 8.7 days — teortaxesTex · 2026-11-05
- Vercel Security Review finds 6,817 vulnerabilities in 24 hours, agent patches 1,411 — cramforce · 2026-11-05