
The thesis: tokens are cheap, verified judgment is scarce
Humans are the constraint that unlocks compute spend; ground truth maps the problem space.
Read the essay →
Essays on measuring what AI can actually take over: the thesis, the method, and what we learn running the loop. New pieces land here first; the short versions go out on LinkedIn.

Humans are the constraint that unlocks compute spend; ground truth maps the problem space.
Read the essay →
In the works

Why unmeasured AI apps stall, and why a demo that works 80% of the time is worthless for work that has to be right.

Judgment loops, agreement rates, the three zones, and diminishing returns as a finding rather than a failure.
Essays on measuring what AI can actually take over. No promotion.