AI tools flooded the design and copy space. The result: a recognizable monoculture. This skill is a systematic antidote. Use it to audit existing work OR to guide new creation from scratch.
How to Use This Skill
- Audit mode — User shares existing UI/copy/UX flow. Run through the diagnostic checklists below. Call out every pattern by name. Suggest replacements.
- Creation mode — User wants new UI/copy/flow. Read the "What to do instead" sections first, then produce work that avoids all listed patterns.
- Spot-fix mode — User points to one specific element. Diagnose it, explain why it's sloppy, rewrite/redesign it.
Always name the sin before fixing it. Specificity builds trust.
Part 1 — AI-Slop UI Patterns (Visual Design)
The Canon of Slop
These visual patterns define the 2024–2026 AI aesthetic monoculture. Flag every one you see:
| Pattern | What it looks like | Why it's slop |
|---|---|---|
| Purple gradient hero | #7c3aed → #2563eb on white bg |
Default Tailwind AI app palette. Seen on 40,000+ products |
| Glassmorphism cards | Frosted glass, backdrop-blur, bg-white/10 |
Overused since iOS 15, now shorthand for "I followed a tutorial" |
| Rounded everything | border-radius: 24px+ on every element |
Removes personality, softens until nothing has weight |
| Inter / DM Sans / Space Grotesk | Default "modern" sans | These three fonts now signal "AI-generated UI" more than any other single cue |
| Emojis as icons in headers | ✨ Supercharge your workflow 🚀 | Startup theater. Hollow optimism. |
| Hero headline formula | [Verb] your [noun] with [product] |
"Supercharge your workflow with Aria" — indistinguishable from 10,000 others |
| Three-column feature grid | Icon + bold label + 1 sentence | Every SaaS landing page since 2019 |
| Testimonial carousel with headshots | Circular avatar, name, company, 1 sentence | Invisible. Nobody reads it. |
| CTA: "Get started for free" | Large button, primary color | Meaningless. Says nothing specific. |
| Empty states with illustration + button | Lottie animation or SVG blob person | Cute once. Now patronizing. |
| Skeleton loaders for everything | Gray pulse bars | Often used to mask poor performance instead of fix it |
| Dark mode: black + purple | #0f0f0f + #8b5cf6 |
The "hacker aesthetic" clone |
| Animated gradient text | Peak 2023 AI startup energy | Looks desperate |
| "Powered by AI" badge | Small badge on the UI | Trust signal that signals nothing |
| Metric cards theater | Big number, small label | Data theater. Not actionable |
What to Do Instead
- Typography first. Choose a font combination that is specific to the context. Research type history. Use a serif with character for body, a grotesque with optical quirks for display — or invert. Never use the font "because it's clean."
- Commit to one extreme. Brutally minimal OR maximally dense. The middle is where slop lives.
- Use real color theory. Complementary pairs, analogous schemes, split-complementary. Not "purple because AI."
- Space is a design element. Generous negative space with one dense anchor beats uniform padding everywhere.
- Let the content shape the layout. Don't force content into a 3-column grid because that's the template.
- Reference actual design movements. Swiss grid. Bauhaus. Emigre magazine. Brutalist web. Dutch constructivism. Tschichold. Pick one and execute it with intent.
Read: references/ui-alternatives.md for specific replacements by component type.
Part 2 — AI-Slop UX Patterns (Interaction & Flow)
| Pattern | Why it's slop |
|---|---|
| Onboarding modal | Interrupts before context. |
| 5-step wizard | Treats users as suspects to be processed. |
| Tooltip tours | Teaches wrong interface instead of fixing it. |
| "Are you sure?" | Trust issues. Use undo instead. |
| Generic empty states | Doesn't explain what item IS or why I'd want it. |
| Toast notifications | Noise. Users learn to ignore them in 2 sessions. |
| Infinite scroll + button | Design indecision shipped as a feature. |
| Exact match search | Punishes the user for trusting the product. |
| 8+ field form | Commitment before value. Backwards. |
| "Loading..." | I don't know if it's 1 second or 1 minute. |
| Action reload | Full page refresh. |
| Hamburger menu | Discovery failure. |
| Hover states only | Mobile, keyboard, discoverers all fail. |
Responsive Anti-Patterns & Best Practices
| Pattern | Slop Cue | Fix |
|---|---|---|
| Hamburger | Hidden nav on desktop | Discovery failure. Use persistent nav. |
| Touch targets | Buttons/links < 44px | Frustrating. Use 44px+ targets. |
| Fixed layout | Horizontal scroll on mobile | Unusable. Use responsive stacks. |
| Hidden actions | Desktop-only visibility | Mobile fail. Keep primary visible. |
| Popups | Hard-to-dismiss modals | Frustrating. Use inline feedback. |
Guidelines:
- Mobile (<640px): Bottom tab bar or drawer; stacked layout.
- Tablet (640-1024px): Collapsible sidebar; 2-column max.
- Desktop (>1024px): Persistent sidebar (280px).
- Verify: ≥44px touch targets; no horizontal scroll; accessible nav.
What to Do Instead
- Don't teach the UI — fix the UI. Tour needed = unclear. Redesign.
- Undo over confirm. Give users a 5–10 second undo window on destructive actions. Way less friction.
- Empty states with one specific next action. Tell users what they'll get, why it matters, exactly what to do.
- Progressive disclosure. Start with the minimum viable form. Add fields only when the user needs them.
- Optimistic UI. Show the outcome immediately, reconcile in the background. Feels instant.
- Contextual notifications. Surface feedback inline, near the action. Not a toast that floats in a corner.
Read: references/ux-alternatives.md for flow-by-flow replacements.
Part 3 — AI-Slop Copy & Text Patterns
This is where AI slop is most pervasive and most invisible. By 2026, entire products are written by language models imitating other language models imitating corporate copywriters. The result is a distinct voice: enthusiastic, hollow, circular, and aggressively affirmative.
The Canon of Slop
| Slop Type | Examples | Fix |
|---|---|---|
| Hollow Affirmations | Absolutely!, Certainly!, Of course! | Delete them. Start with content. |
| AI Superlatives | Powerful, seamless, intuitive, robust | Use specific claims and data. |
The Listicle Reflex
AI defaults to bullet points for everything. Three bullets where one sentence works. Numbered lists for concepts with no actual sequence. Fake hierarchy with **Bold:** then explanation for everything.
Fix: Write prose. Use a list only when there is a genuine enumerable set AND scanning adds value.
Transition Theater
- "In conclusion...", "To summarize...", "In essence..." / "It's worth noting that...", "It's important to remember..."
- "With that said...", "Having said that..." / "At the end of the day..."
Fix: Just say the thing. These phrases delay the idea without adding to it.
Emoji Inflation
Using 🚀 💡 ✨ ⚡ 🔥 as substitutes for meaning. One emoji in a headline was interesting in 2020. Now it's punctuation-as-performance.
Fix: Use zero emojis unless the context is genuinely casual/social. If you use one, mean it.
The Feature Announcement Formula
"We're excited to announce [Feature]! This powerful new capability lets you [vague verb] your [noun] like never before. Stay tuned for more updates!"
Fix: What does it do, concretely? What problem did it solve? Who asked for it? Write a changelog, not a press release for your own team.
Hedging Chains
- "This may potentially be a possible consideration..." / "Generally speaking, in most cases..." / "You might want to consider..."
Fix: Own your statements or explicitly flag uncertainty once, clearly.
Product Copy Sins (Landing Pages)
| Sin | Example | Fix |
|---|---|---|
| Features listed as verbs | "Collaborate, Create, Ship" | What does it actually DO? |
| Social proof without specifics | "Trusted by 10,000+ teams" | 10,000 teams doing what? What outcome? |
| Vague problem statement | "Work is broken" | Whose work? Broken how? |
| The "journey" metaphor | "Begin your journey today" | It's software, not Tolkien |
| Future-tense promises | "Will change the way you think about X" | Show it changing it. Now. |
UX Writing Sins
| Sin | Example | Fix |
|---|---|---|
| Error: blame the user | "Invalid input" | "Email addresses need an @ sign" |
| CTA: describe the UI action | "Click here" | "Download the report" |
| Label: use noun for verb slot | "Settings" button that saves | "Save settings" |
| Success: announce the action | "Saved!" | "Changes saved — live in 30 seconds" |
| Placeholder as label | Input with placeholder "Email" and no label | Use a real label. Always. |
| Confirmation copy that restates the question | "Are you sure you want to delete? This will delete the item." | "Delete [Item Name]? This can't be undone." |
Code Slop Patterns (Internal Implementation)
Detect and fix these generic, low-quality patterns in AI-generated code:
| Pattern | Example | Fix |
|---|---|---|
| Vague placeholders | // TODO: implement this |
Implement the logic or remove the comment if redundant. |
| Generic variables | data, result, temp, item |
Use descriptive names: userProfile, calculationResult. |
| Valueless boilerplate | Redundant getters/setters in languages that don't need them. | Use language idiomatic features (e.g., TS properties). |
| Unnecessary verbosity | Nested if chains where a guard clause works. |
Refactor with guard clauses and early returns. |
| Commented-out code | Large blocks of unused code. | Delete them. Git is your history. |
Part 4 — Audit Workflow
When given something to review:
- Scan for patterns. Check all three canons (UI, UX, Copy). List every match by name.
- Score severity. Not all slop is equal:
- 🔴 Structural — Requires redesign or rewrite. Fundamental problem.
- 🟡 Surface — Easy fix. Wrong word, wrong color, wrong font.
- 🟢 Cosmetic — Minor. Polish pass.
- Prioritize. Fix structural first. Don't polish a broken foundation.
- Rewrite/redesign. For each flagged item, provide the specific replacement — not generic advice.
- Explain the why. Name the design principle behind each fix.
Part 5 — The Positive Doctrine
Anti-slop isn't just negation. These are the affirmative principles:
Design
- Specificity > universality. Design for this user, this task, this moment.
- Tension is interest. Contrast, asymmetry, and friction (used deliberately) are memorable. Harmony can be invisible.
- Constraints create identity. Impose a real restriction and design within it. The best brands have rules.
- Reference the real world. Materials, textures, physical objects, historical artifacts. Not just other apps.
UX
- Respect the user's time. Every click, form field, and modal is a tax. Minimize it.
- Be opinionated. Show users the best path. Don't present 6 equal options when one is clearly right.
- Context over consistency. The right interaction for this moment > the component library default.
Copy
- Specific > general. "Saves 3 hours per week" > "Saves time"
- Active > passive. "We deleted it" > "It was deleted"
- Short > long. Read the sentence aloud. Cut every word that doesn't earn its place.
- Voice = point of view. Have one. Don't round off all the edges trying to please everyone.
- Write for one person. Not "users". Not "teams". The specific human who will read this specific sentence.
Rationalizations
| Rationalization | Reality |
|---|---|
| "Users are used to this kind of UI anyway" | Familiarity with mediocrity isn't an excuse to propagate it. |
| "It's faster to use the default AI-generated copy" | Fast slop is still slop. Quality requires human-like consideration and editing. |
| "This doesn't need to be unique, it just needs to work" | Distinctiveness is part of 'working'. It builds brand trust and memorability. |
Red Flags
- Over-reliance on "modern" sans-serif fonts without stylistic justification
- Using generic affirmations like "Absolutely!" in UX copy
- Defaulting to 3-column feature grids for all content types
Reference Files
references/ui-alternatives.md— Component-by-component replacements for slop UIreferences/ux-alternatives.md— Flow-by-flow replacements for slop UX patternsreferences/copy-rewrites.md— Before/after rewrites for the most common AI copy sinsreferences/inspiration.md— Design references, typographers, movements to study
Read these when you need specific replacements or need to justify a creative direction to the user.