The levels of agentic engineering: find what you still do by hand, then automate it
rseroter · x · 2026-11-18
Zach Wills, who has twice led AI transformation (most recently for a 100-person R&D team in a 700-person company), argues that "levels of agentic engineering" lists read like tool ladders. The better frame: one question — what are you still doing yourself? Every level is removing yourself from the next part of the loop.
Key points:
- Ship, don't mandate: demonstrate the faster way by delivering real software alongside the team
- Shipping surfaces systemic friction: infrastructure, review and compliance policies that block automation only appear when you actually ship
- His Level 6 diagram shows a multiplayer factory: product/support feed ideas and bugs into build/test/QA agent stations with shared e2e/QA/arch checks
More from coding & agent
- Higgsfield's new AI film: 'The hardest part is making it not look like AI' — CurieuxExplorer · 2026-11-18
- Project triangle flips in the AI agent era: defining 'good' becomes the job — lak_luster · 2026-11-18
- GitHub Next's Maggie Appleton on speculative design for prototyping software futures — JayAlammar · 2026-11-18
- SWE-CoWork benchmark: best model fully passes only 8 of 101 handoff coding tasks — omarsar0 · 2026-11-18
- LangChain explains Agent vs. Deep Agent vs. Managed Deep Agent — LangChain · 2026-11-18
- Dev doubts extreme multi-agent workflows: little value, multiple subscriptions burned — DanielLockyer · 2026-11-17