Writing
Technical notes on AI system boundaries, project truth, browser text layout, and the work required to keep active builds honest.
Dispatches / 01
Articles and notes
Five records on system boundaries, project truth, and the mechanics behind the work.
Designing Decision Boundaries for AI Agents: Clarence as an Agentic UX Case Study
How Clarence defines what an AI collaborator can decide, what it must verify, and where human judgment stays non-negotiable.
Read article →Working With Clarence
How I use Clarence for UX critique, creative prototyping, and system workflow without handing off judgment.
Read article →Why Clarence Needed a System Wiki
Why important system knowledge could not stay trapped in chat logs, repo docs, and a vault that could point at the wrong place.
Read article →Why SensorSynthFM Needed a Wiki Before It Needed More Features
Why SensorSynthFM had a truth problem, not an idea problem, and how a project wiki keeps the thesis-demo work honest.
Read article →Implementing Pretext for Character-Level Text Layout
How I used Cheng Lou's new 15KB library to replace DOM-based text measurement with Canvas-powered character-level layout, and why it matters for interaction design.
Read article →