Taste Engineering · 8 / 12

Generating UI: From Adjectives to a Spec

When you brief AI on UI, adjectives are wishes and a spec is an instruction. Same landing-page request, three prompt tiers, three quality tiers of output. This lesson teaches you to write “looks good” as a spec AI can't wander from.

Three-tier compareSpecTypography specsNegative constraintsPrompt blocks
Adjectives are wishes; a spec is an instruction

Hand over “make a nice landing page,” and AI hands back purple gradients and centered filler — generic stuff. AI only owns half the blame: with no information, it can only output the training-data average. Prompt engineering already said “you get what you put in”; generating UI is the clearest demo of that rule. Lock the primary color, pin spacing to a grid, limit weights to two levels, list the bans — and the output jumps a tier.

Try it · Same brief, three prompt tiers
Three prompt tiers, three outputs Switch tiers
Flip between them a few times — feel how prompt information turns into output quality.
PromptMake a nice landing page
untitled-project.vercel.app
AI-powered · Next-gen platform
Redefining the future of productivity
All-in-one · Every scenario · Empower every team
Get startedLearn more
The choice of 100,000+ teams
Smart
AI-powered
Efficient
Blazing fast
Secure
Enterprise
This is the default “AI look.” With no information, AI outputs the training average: purple gradients, glass cards, centered filler, slogan copy. It looks like every AI-generated page — and nothing like your product.
Typography has hard numbers — just copy them

The easiest thing to miss in a spec is the text itself. Interfaces are mostly type — if type fails, perfect color won't save you. Good news: typography already has hard numbers. About Face 4, Chapter 17, hands you the figures: contrast, typeface, case, size — all four translate straight into prompts.

Contrast at 80%Body-to-background luminance contrast at least 80%. Light gray on light gray may feel “premium”; to users it's standing at attention to read.
Sans-serif on screenSerifs fray on screens — body text is always sans-serif; serif only on large decorative titles.
No all-caps EnglishPeople recognize word shapes; all-caps flattens them — loud and hard to read. Build hierarchy with size and weight, not by shouting.
10px is the floorBelow 10px, type is basically unreadable. Even disclaimers must be readable — smaller and they're just decoration.

Mullet and Sano's verdict, quoted as-is in About Face 4: unnecessary variation is the enemy of consistent design. Things that are nearly the same should be the same. Write it into the prompt — it's the highest-value ban.

Try it · Spec translator

Left: how you talk. Right: the spec AI needs. Tap a row to translate it; finish all five and copy the set.

Spec translator 0 / 5
Tap a fuzzy ask on the left to reveal a numeric spec on the right; finish all five and copy in one click.
“Make the type readable”
Tap to translate
“Tone down the headlines”
Tap to translate
“Easy to read on screen”
Tap to translate
“Don't make it flashy”
Tap to translate
“Make it look more professional”
Tap to translate
Try it · Typography spot-the-diff
Same card, two typography sets Pick one
Tap the one that reads better, then check it against the four rules above.
A
teamly.app/pro
PREMIUM PLAN LIMITED OFFER
Upgrade to Pro for advanced reports and team collaboration. Order this week for 20% off.
Platform reserves final interpretation. Renewals bill at full price. See terms.
B
teamly.app/pro
Pro — limited offer
Upgrade to Pro for advanced reports and team collaboration. Order this week for 20% off.
Note: renewals bill at full price. See terms.
Try it · Build your own spec

You don't memorize a spec — pick one block from each of six clauses and it's assembled. Wherever you choose “leave blank,” that part goes back to AI luck — and the preview shows it clearly.

Prompt block builder Live assemble
Pick one block per group; the prompt below updates. When it looks right, copy and send to AI.
Reference product
Primary
Spacing
Weight
Type
Bans
One more quiz before you go
Scenario: which constraint is worth most Single choice

Which constraint lifts AI-generated UI the most?

AMake it premium and elevated
BColors should be harmonious
CNo gradients; only one primary color plus neutral grays
DReference big-tech style
Three clause types — copy these lines
Clause typeRoleTemplate wording
Positive specsLock everything you can quantify so AI has nothing to guess“Primary #DB2777, spacing on an 8pt grid, corner radius 12px everywhere, weights only 800 and 600”
Negative constraintsBlock the paths AI loves to wander down“No gradients, no emoji icons, no more than one shadow style”
Reference anchorUse a product both sides know as the coordinate“Information density reference Linear; whitespace reference Notion.” Don't say “big-tech style” — no pointer means you said nothing
Unity clauseUnify “nearly the same” into the same, ahead of time“Make near-identical elements identical: one primary, one radius set, one shadow style sitewide; delete unjustified variation”

The fourth type is Mullet & Sano's “unnecessary variation is the enemy of consistent design,” rewritten as a prompt. AI loves to improvise details: 10px radius here, 12px there, two grays, three shadows. A unity clause seals that gap early.

Key Takeaways

Adjectives are wishes; a spec is an instruction: primary, spacing, weight, bans — written as clauses you can verify one by one.

Negative constraints give the best ROI: “No gradients, no emoji icons, no more than one shadow” — three lines cut most of the AI look.

Typography has hard numbers to copy: 80% contrast, sans-serif body, no all-caps English, 10px minimum — four acceptance lines in the prompt.

Next time you generate UI, assemble a spec with the block builder and compare the output to last time's offhand one-liner.

Source: Original to Xiaoshan Academy's Taste Engineering series; some design principles adapted from About Face 4: The Essentials of Interaction Design, Chapter 17.