Writing
What I think about when I build.
Essays in my own words about applied AI, operating judgment, evidence, model behavior, and the kind of capability technology should create.
How I work
Proof before promise
I would rather describe what is real, what is still uncertain, and what the evidence actually supports than let ambitious language outrun the work.
January 16, 2026 / 6 min read
Applied AI
Why operator experience makes better AI products
The best AI products begin where a real system already has stakes: a physical workflow, a data workflow, a release process, or a product surface where mistakes matter.
February 13, 2026 / 9 min read
Applied AI
General AI gets useful through real architecture
AI's generality is the breakthrough. The product work is building the harness, memory, tools, evaluations, and context that let that generality do real work.
March 20, 2026 / 8 min read
Model evaluation
If models matter, measurement matters
AI Psychometrics Lab makes an important point: we should not only use models, we should measure how they behave.
April 24, 2026 / 9 min read
Worldview
Technology should create capability, not just novelty
The deeper thesis is that technology should expand human capability in ways that are concrete, durable, and worth caring about.
May 22, 2026 / 8 min read