Δ Delta
From the book · free at acuity.press

The Ten Laws of
Enterprise AI Prompting

Structural truths about AI-assisted development — independent of model choice, stable across versions. Click any law for the short version; read the book for the working code.

Law 1. The AI produces what is average in its training data. Your domain requires what is specific. Specificity must come from you.

Law 2. A vague specification is not a prompt. It is a request for the model's best guess. Only you know if the guess is right.

Law 3. The quality of your output is bounded by the quality of your constraint. Missing one constraint leaves one failure mode open.

Law 4. Context not provided is context invented. The model fills every gap — just not with your actual codebase.

Law 5. No float for money. No exceptions. No ‘it's fine for this use case.’ No.

Law 6. Every public-facing AI feature is an injection surface until proven otherwise.

Law 7. The best debugging prompt is a complete Evidence Brief. Fill it before you start.

Law 8. Prompts are code. Code that is not version-controlled, reviewed, and tested will degrade.

Law 9. An agent given a vague specification does not produce one bug. It produces a system of bugs.

Law 10. The Specification Gap never closes itself. Only deliberate engineering practice closes it.

Read the full book — freeScore your prompt against the lawsCompanion code (MIT)