Writing.
Short essays and field notes from building with LLMs. Mostly on prompts, agents, evals, and the quieter UX problems that come with putting models into real workflows. Published weekly-ish.
2026 · 4 posts
2026 · 04 · 05
Cost and amnesia: the two quiet failures of cloud assistants.
Agents
6 min
2026 · 03 · 14
If the rulebook already exists, don't ask the model to re-learn it.
Practice
5 min
2026 · 02 · 28
When "all green" lies: monitoring only what you can see.
Engineering
6 min
2026 · 01 · 15
Broad before deep: why I'm building a bit of everything.
Essay
4 min