Direct answer
A source-of-truth layer is the locked part of a knowledge pack that contains verified facts, definitions, boundaries, examples, caveats and exclusions. It tells the AI what must not drift when producing summaries, articles, prompts, social posts or strategic recommendations.
Why this matters
A source-of-truth page gives the system a practical anchor. It also lets later articles explain hallucination prevention without repeating the whole Choice Engine method.
How to use it
- Create a locked section for definitions, claims and exclusions.
- Put examples and caveats next to the claims they qualify.
- Mark anything not yet verified as an assumption or research question.
- Keep delivery formats outside the locked layer.
- Review every derived output against the locked layer.
Example
The definition of bounded divergence stays fixed. A LinkedIn post, article, checklist or prompt may vary in length and style, but it must not change the definition, proof standard or caveat.
Caveats and limits
A source-of-truth layer is only useful if it is maintained. Old facts, stale examples or unreviewed assumptions can become a new source of error.