The Visible Crack
Design Ops
Product Design
When AI generation enters a multi-brand design workflow, the first thing it does is show you exactly how well your system was actually built.
Clean token inheritance - core down to brand layer down to product - means AI generates correctly across every brand. Inconsistent naming, hard-coded brand logic, governance that was documented once and never maintained - all of that produces off-brand output simultaneously, at volume, before anyone catches it.
At a single-brand scale, that kind of debt stays manageable. One team drifts from the naming convention, one component gets rebuilt instead of reused - it surfaces slowly, gets fixed in the next sprint. Add AI generation across six brands and the same drift runs in parallel across all of them. The pace changes what the problem is.
Multi-brand design at scale has always required a level of architectural discipline that single-brand work can defer. AI generation just removed the option to defer it.
As auditors go, AI is unusually honest. It doesn't file a report. It just generates - and what comes out tells you everything about what went in.