AI · Brand systems In development
Machine-led branding standards.
Brand guidelines were written for humans — a PDF a designer reads, interprets, and applies with judgment. AI doesn't read a playbook that way. So I'm rebuilding one that a machine can actually follow.
The premise: as more creative gets generated rather than made, the brand playbook has to become a system of rules, tokens, and examples precise enough for a model to execute — and clear enough that a human still owns the taste. This project takes a traditional brand playbook and rewrites it for that world.
What it will cover
Tokenized identity
Color, type, spacing, and logo rules expressed as machine-readable tokens, not just pages — so a generator can't drift off-brand.
Voice, as constraints
Tone and copy standards written as rules and worked examples a language model can apply consistently across formats.
Prompt & guardrail library
Reusable prompts, do/don't examples, and guardrails that keep generated creative inside the brand — with a human check at the seams.
The human line
An explicit map of where the machine executes and where judgment stays — the part of the standard that protects the craft.