AI in a Design Workflow: What We Use It For (and What We Don't)

Every agency now claims “AI-powered workflows.” Here’s what that concretely means in ours — and the places we deliberately keep it out.

Where AI genuinely accelerates the work

  • First drafts of structure. Sitemaps, content outlines, user-flow variations — AI produces starting points in minutes that used to take a morning. We keep maybe 40%, but the 40% arrives instantly.
  • Development speed. Boilerplate, refactors, and test scaffolding move dramatically faster. The craft shifts to review: knowing what good looks like so you catch what’s off.
  • Research compression. Summarizing competitor positioning, extracting themes from customer reviews, auditing a site’s technical SEO — hours to minutes.
  • Asset drudgery. Alt text drafts, image resizing, metadata passes — automated entirely.

Where we keep humans on the wheel

  • Taste. AI converges on the average of everything it’s seen — which is exactly what a brand paying for distinctive work doesn’t want. Direction, art direction, and the “this feels wrong” instinct stay human.
  • Strategy. Knowing which problem to solve. AI answers the question you ask; the expensive skill is asking the right one.
  • Client copy. Drafting assistance, yes. Publishing unedited AI text under a brand’s voice, no — readers notice, and increasingly, so does search.

The tell of AI done badly

Sameness. Generic gradient heroes, interchangeable copy (“Unlock your potential with seamless solutions”), a hundred thin blog posts published in a week. Google’s spam policies now specifically target scaled low-value content — the shortcut is also the penalty.

Our position

AI compresses production; it doesn’t compress judgment. Studios that use it well ship better work faster. Studios that use it as the product ship faster mush. The difference shows in the portfolio — judge ours.

Next project

Have an ideaworth raising?