Start here

TL;DR

This site explains how AI answers behave and stabilize, based on observable patterns.
We focus on belief states: what’s confident, what’s hedged, and what’s missing.
We don’t claim internal model mechanics. We document repeatable outputs.

What this is

Field guide: A set of concepts and templates to map AI belief, human belief, and the overlap of trusted sources — then publish scaffolding that makes understanding more coherent.

What this is not

Not a “rank for AI Overviews” playbook.
Not internal ML architecture documentation.
Not a collection of prompt hacks.

The north star

We don’t optimize for AI outputs. We influence the belief systems AI reflects.

How to use this

Map AI belief: run prompts, capture hedges, capture citations.
Map human belief: forums, social listening, recurring confusion.
Find gaps/tensions: where AI is confident but humans aren’t (and vice versa).
Publish scaffolding: definitions → mechanism → constraints → application.
Re-test: does AI hedge less? does it adopt the framing?

Last updated: 2026-02-04