From Prompt to Token: How an LLM Generates a Response
How a causal language model turns a prompt into the next token: tokenization, layers, logits, sampling, and the cache that makes the loop possible.
Ideas
Questions, tests, and unfinished thinking
This is where I explore questions, test ideas, learn in public, and invite other people into unfinished thinking. A status of Active, Exploring, Concept, Incubating, or Paused means what it says.
What a measure needs before anyone should trust it
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.
Independent Practice Intelligence
A venture inside CincyLogic, founded August 2026. Assessment-led practice intelligence for independent specialty and community practices. The public entry is the Independent Practice Resilience Assessment.
CHASM
An experimental data and analytics systems project. A public description of the mechanism is not ready.
AI research lab
Public experiments in understanding and explaining language-model systems.
Workflow and product concepts
A held place for scheduling systems, process tools, and diagrams that are not products yet.
Nothing in this type yet.