Field notes · Essays · Working ideas
Notes on LLMs, AI, and the industries they are reshaping — collected from the field by a Forward Deployed Engineer who writes things down so he doesn't lose them.
Syngha Han · MMXXVI
Why AI transformation in legacy enterprises is fundamentally a translation problem between AI engineers' timelines and operators' timelines — and why the Forward Deployed Engineer role is becoming the most important translator we have.
AI's productivity gain doesn't lower development cost. Why most layoffs blamed on AI weren't, what LLMs actually accelerate, and why cost-efficiency framings of AX are aimed at the wrong target.
AI is still a technology whose shape changes by the week. Memorizing textbook definitions or this quarter's tool manuals is mostly wasted effort. Following the history — what got built since ChatGPT, what each lab was looking toward when they built it — is the coordinate system that lasts. An interactive timeline from BERT (2018) to Claude Opus 4.7 (2026).