← @bgSponge

How to scale intent, quality, and artistry with Al | Katie Dill (Stripe)

Summary

Katie Dill argues that designers must encode brand intent into machine-readable systems to prevent AI from producing generic,

  1. LLMs default to generic patterns.

    Because LLMs predict the most probable answer, they replicate existing styles rather than creating brand-specific or context-aware designs.

    LLMs are really good at telling you the most probable answer

  2. Speed creates the burrito dilemma.

    The rapid execution of AI mimics a microwave burrito, where users accept mediocre,

    I'm willing to overlook some pretty serious flaws. It's nearly inedible, but I'm so enamored by the speed of execution.

  3. Systems must scale intent, not consistency.

    Stripe uses a CLI built on design systems to make AI more obedient, moving beyond atomic components to full, opinionated templates.

    the old system scaled consistency, but the new system needs to scale intent.

  4. Editors must prevent

    Since AI allows building 20 ideas in a week, organizations need human editors to perform post-build filtering to ensure quality.

    the filtering that used to be throughout the process now needs to happen post-build

  5. AI acts as a creative catalyst.

    Rather than just automating existing tasks, AI can enable complex iterations, such as a designer performing 56 passes on an animation.

    AI can help us raise the ceiling, not just the floor.

Read the original ↗Save to your Sponge

Summarized with Sponge. Quotes are the author’s own words.

More from @bg

All readings →