Encoding Brand Rules

Brand guidelines written for humans are aspirational. Brand rules encoded for machines are enforceable. The translation from one to the other is the foundation of AI brand enforcement.

Human guideline: "Use the brand cyan as the primary accent color." Machine rule: "Primary accent must be #00ffff. Acceptable deviation: hue +-2 degrees, saturation +-5%, lightness +-3%. Secondary accent #ff9f1c may appear at no more than 15% of total color area. Any color not in the approved palette (brand.json colorSystem) triggers a violation."

Every brand guideline translates into one or more machine rules with four components. The Property: what visual element is governed (color, font, spacing, composition). The Value: the exact specification (hex code, pixel value, font name, ratio). The Tolerance: acceptable deviation range. The Violation Response: what happens when the rule is broken (reject, flag for review, auto-correct). Vague guidelines that cannot be translated into this four-component structure are not enforceable — and unenforced guidelines are suggestions that erode over time.

1. Identify the Property What visual element does this guideline govern? Color, typography, spacing, composition, imagery style, effect application. Be specific — "color" is a category; "primary accent color" is a property.
2. Define the Value and Tolerance What is the exact specification? How much deviation is acceptable? #00ffff with +-2 degree hue tolerance. Exo 2 Bold with no font substitution. 20px minimum whitespace with +-4px tolerance. Values without tolerances are either absolute or undefined — both are bad.
3. Set the Violation Response Reject: the asset is blocked from approval. Flag: the asset is approved with a warning for human review. Auto-correct: the system adjusts the violation automatically (color shifting, font substitution). Choose the response based on violation severity — wrong brand color is a reject; 2px whitespace deviation is an auto-correct.