Research
Not a blog. Notes with a type, a depth, and a status. The simplest useful version comes first.
Explainer
How a causal language model turns a prompt into the next token: tokenization, layers, logits, sampling, and the cache that makes the loop possible.
Executive / Practitioner / Technical
Updated 2026-10-04
Explainer
Grain, time, inclusions, an owner, and a known way to be wrong. A teaching note on KPI design with a labeled example and no client metrics.
Executive / Practitioner / Technical
Updated 2026-10-04
No notes in this topic yet. The shelf is real; the piece is not written.