Anti-AI Design Slop

AI can now generate a decent-looking visual in seconds.

That is exactly why visual judgment matters more.

Most AI-generated design fails in the same way: it looks polished, but it says nothing specific.

This repo gives you a practical checklist, scoring model, and prompt pack for rejecting generic AI-generated visuals before they reach your website, deck, or LinkedIn feed.

If this helps you improve AI-generated visuals, please consider starring the repo so others can find it.

Use this in 5 minutes

1. Open STARTER-PACK.md.
2. Copy the system instruction into your AI tool.
3. Paste your visual prompt, image description, carousel draft, website hero, or slide concept.
4. Ask the tool to score it from 0 to 5.
5. Reject anything below L4 for public content.
6. Use prompts/prompt-rewriter.md to improve weak prompts.
7. Run prompts/final-qa-gate.md before publishing.

What this is not

This is not anti-AI.

This is not a claim that gradients, motion, 3D visuals, or visual trends are always bad.

This is a review system for checking whether a design choice has a job.

A visual effect is fine when it improves clarity, trust, or memory.

It becomes slop when it only fills space.

What is AI design slop?

AI design slop is visual output that looks professional at first glance but fails under review.

It often has:

• no clear message
• no ownable brand system
• too many effects
• weak hierarchy
• text that is hard to read
• visuals copied from common AI SaaS patterns
• decoration that does not help the viewer understand anything

The result is design that feels modern but not memorable.

Who this is for

This repo is for:

• founders
• marketers
• designers
• content creators
• operators
• product teams
• AI agent builders
• anyone using AI to create visuals, websites, decks, or social content

You do not need to be a designer to use it.

You do need to care about clarity, originality, and trust.

The core idea

Do not approve visuals because they look polished.

Approve visuals only when they are:

1. original
2. clear
3. brand-specific
4. readable
5. credible
6. useful

If a design could belong to any AI startup, reject it.

What this repo includes

anti-ai-design-slop/
├── README.md
├── STARTER-PACK.md
├── LICENSE
├── .gitignore
├── checklists/
│   ├── universal-checklist.md
│   ├── accessibility-checklist.md
│   ├── linkedin-carousel-checklist.md
│   ├── marketing-graphics-checklist.md
│   ├── executive-deck-checklist.md
│   └── website-hero-checklist.md
├── prompts/
│   ├── visual-storytelling-agent.md
│   ├── anti-slop-reviewer.md
│   ├── prompt-rewriter.md
│   ├── linkedin-carousel-generator.md
│   └── final-qa-gate.md
├── scoring/
│   ├── maturity-model.md
│   ├── dimension-rubric.md
│   └── pass-fail-gate.md
├── examples/
│   ├── bad-vs-good.md
│   ├── before-after-reviews.md
│   └── review-template.md
└── docs/
    └── public-use-warning.md

Start here

If you only use one file, start with:

STARTER-PACK.md

If you want the full review checklist, use:

checklists/universal-checklist.md

If you are creating LinkedIn content, use:

checklists/linkedin-carousel-checklist.md

If you need accessibility checks, use:

checklists/accessibility-checklist.md

If you are building an AI visual agent, use:

prompts/visual-storytelling-agent.md

If you want a strict critic, use:

prompts/anti-slop-reviewer.md

Fast review method

Before publishing any AI-generated visual, ask:

1. Can the target viewer understand the message in 5 seconds?
2. Would this still feel specific if the logo was removed?
3. Does every visual element have a job?
4. Is the text readable on mobile?
5. Would a serious buyer trust this?
6. Does it look like generic AI SaaS?

If the answer to question 6 is yes, revise or reject.

Pass / fail gate

Pass only if:

• Originality score is 4 or higher
• Message clarity score is 3 or higher
• Hierarchy score is 3 or higher
• Readability score is 3 or higher
• Buyer trust score is 3 or higher
• Brand specificity score is 3 or higher

Reject if:

• it looks like generic AI SaaS
• it depends on decoration instead of message
• it hides key information
• it uses motion without purpose
• it could belong to any startup
• text is not readable on mobile

Design maturity model

Level	Name	Meaning	Decision
L1	AI Slop	Generic, decorative, low-trust	Reject
L2	Generic SaaS	Professional, but not ownable	Reject for public work
L3	Professional	Clear, but not distinctive	Use for internal drafts
L4	Premium B2B	Credible, clear, brand-aligned	Publishable
L5	Iconic Visual System	Ownable, memorable, repeatable	Use as reference standard

For scoring detail, use scoring/dimension-rubric.md.

Example: weak vs stronger prompt

Weak prompt

Create a futuristic AI landing page with purple gradients, glowing networks, and floating glass cards.

Stronger prompt

Create a clear B2B landing page hero for a technical buyer.
Use one specific visual metaphor that explains the product's value.
Avoid generic AI SaaS visuals, glowing blobs, abstract neural networks, and decorative gradients.
Prioritize message clarity, trust, readable typography, and a strong first-screen hierarchy.
Before finalizing, score the concept for originality, clarity, hierarchy, brand specificity, readability, and buyer trust.
Reject the concept if it could belong to any AI startup.

More examples are in examples/before-after-reviews.md.

Recommended workflow

Generate concept
→ Run anti-slop reviewer
→ Score with maturity model
→ Rewrite prompt if needed
→ Revise or reject
→ Run final QA gate
→ Publish only if it passes

Public use warning

Do not publish private brand strategy, customer names, internal positioning, or confidential prompt instructions in a public fork.

Keep company-specific visual rules in a private repo or internal agent file.

See docs/public-use-warning.md.

Inspiration

This framework was inspired by a practical problem: AI tools make it easier to create polished visuals, but harder to maintain originality, clarity, and judgment.

It is not tied to one tool or model. It is designed to work with any AI workflow that generates or reviews visual concepts.

Suggested repo topics

Use these topics on GitHub:

ai-design
prompt-engineering
visual-design
design-review
linkedin-carousel
marketing-design
ai-agents
b2b-marketing
brand-strategy
creative-ops

Suggested repo description

A practical checklist and prompt pack for rejecting generic AI-generated visuals before they become design slop.

Suggested social post

I turned my AI design review process into a public repo.

Anti-AI Design Slop is a checklist and prompt pack for catching generic AI visuals before they get published.

It helps review:
- LinkedIn carousels
- marketing graphics
- website heroes
- executive slides
- AI-generated visual concepts

The goal is simple:
Stop approving visuals just because they look polished.
Approve them only when they are clear, original, readable, and brand-specific.

License

MIT.

Contributing

Ideas, examples, and review templates are welcome.

Good additions include:

• before/after examples
• stronger design prompts
• industry-specific anti-slop checklists
• visual QA templates
• agent instruction files

Maintainer note

This is a practical operating system for visual judgment.

AI can generate options.

You still need taste, clarity, and standards.