Author’s Note: Substack isn’t letting me invite bylines at production time, so I’m tagging the remaining contributors here. Huge thank you to:

Some AI-built brands look like stock photo factories.

Others look like high-end studios with a signature.

The difference isn’t talent. It’s not budget either. It’s approach.

You’ve generated hundreds of images by now. You know AI can produce impressive output. The tools keep getting better, the renders keep getting sharper, and yet—something’s still off. Your brand doesn’t feel like yours.

It feels like it could belong to anyone with the same subscription.

What this missing skill is costing you

The problem isn’t AI’s capability. You already know the tools can produce stunning visuals. The problem is the art direction layer—knowing how to lead the tool rather than accept its defaults.

Most creators treat AI like a slot machine. Type a prompt, pull the lever, hope for something good. When it doesn’t land, they regenerate. And regenerate. And regenerate. Hours vanish into that loop without progress.

Meanwhile, their brand presence shifts with every post. Monday’s aesthetic doesn’t match Friday’s. The color palette drifts. The style wanders. Audiences can’t recognize them because there’s nothing consistent to recognize. Time lost and brand equity eroded simultaneously.

When Coca-Cola released their AI-generated “Holidays Are Coming” ads in late 2024, the backlash was immediate and brutal. Viewers called the visuals “soulless” and “creepy.” Even massive brands can’t escape this when they go light on the art direction layer and instead let the “AI-ness” of it take the lead.

Coca-Cola was unscathed because they’re a multi-billion dollar company. Your business might not have that cushion.

So I surveyed 10 creators with standout visual brands

Some creators have figured this out. Their AI-assisted brands look distinctive, consistent, recognizably theirs. I wanted to know what they were doing differently.

I asked ten AI creators who’ve cracked visual branding three questions:

• What’s one visual branding habit you swear by?

• What mistake would you warn others to avoid?

• And what’s the difference between a brand that looks AI-generated and one that doesn’t?

Then I synthesized their answers. Three core distinctions emerged, along with seven specific habits that separate creators who blend into the AI noise from those who stand apart.

The three core distinctions

Across every response, three themes surfaced again and again.

• Human Direction. You’re the art director, not the passenger. The AI proposes; you dispose. This includes both vision-setting—knowing what you want before you prompt—and refinement after. As Daria Cupareanu put it, “strong brands treat AI output with the same scrutiny and refinement as any other creative tool.”

• Visual System. You’re building a world, not generating isolated images. Locked elements create recognition. Fluid elements create surprise. Consistency without monotony.

• Artistic Soul. Each piece is craft, not content production. Whatever image you create, whatever marketing material, the core of standout visuals remain the same: art. That means imperfection, emotional rexture, and carefully-executed pattern disruption. Mia Kiraki described the goal as “beautiful messiness,” a recognizable style that still surprises.

This aligns with broader research on brand recognition. A Lucidpress study found that consistent brand presentation across all platforms can boost revenue by up to 23%. Visual systems work—whether AI assists in building them or not. We just need to learn using AI to assist in building visual systems well to build recognition.

This also aligns with my broader concept of building irreplaceability. Human direction, a repeatable system, and a human soul is core to building real moats that make your brand Unpromptable.

The 7 habits of creators who don’t look AI-generated

Those three distinctions are the philosophy.

These seven habits are how creators live it daily.

1. Act as the art director, not the passenger

Successful creators don’t open a prompt box and see what happens. They arrive with a vision already formed.

Jake Handy insists you need a clear idea before touching the tool—letting models decide too much leads to generic graphics that scream “AI.” Anna Levitt takes this further: she never generates images until she’s finished writing the article. She reads the draft, identifies the emotional moments, sketches 3-4 concepts by hand, then prompts.

This is the posture shift that separates amateurs from professionals. The amateur asks AI to create something interesting. The professional treats AI like a cinematographer who needs specific instructions—lighting, composition, mood, purpose.

Pinkie captured it precisely: “When thought leads and AI follows, the work reads as human. When AI leads, it shows.” The tool should execute your vision, not generate one for you.

2. Establish a “locked vs. fluid” infrastructure

Every creator I surveyed mentioned some version of this framework.

Mia establishes 3-4 non-negotiable visual rules before generating anything—a specific color palette, consistent lighting style, a “film” genre, and a “film stock” texture. Pinkie treats visual rules as “non-negotiable infrastructure, not creative preference.” The locked elements are what make your brand recognizable at a glance.

But you can go too far.

Some creators fail because they lock everything, and their brand becomes “wallpaper.” Nick Quick warned about this trap—“same aesthetic, slight variations, and eventually it all becomes invisible. Nothing stands out because everything looks the same.” The solution is strategic variation within constraints.

Dallas varies weather, scale, and character activities. Anna experiments with paper textures. The locked creates recognition; the fluid prevents “emotional flattening” where every post feels identical and audiences tune out.

3. Force creative aesthetics and intentional imperfection

AI defaults to hyper-realistic, shiny, digital.

Mia, who trains AI models professionally, lists the dead giveaways: skin textures that look airbrushed, light sources coming from three different directions, the standard blue-and-orange cinematic lighting, compositions that feel too perfectly balanced. Human art has what she calls “beautiful messiness.” AI output rarely does, unless you force it.

Anna Levitt forces AI into “analog” mediumsl; rough pencil lines, watercolor bleeds, ink sketch styles inspired by The New Yorker. She prompts for imperfection deliberately: subtle grey watercolor bleeds, specific color palettes (blue, purple, cyan because she loves Caribbean waters), plenty of white space.

Dallas focuses on “medium behaviour.” How transparent washes bleed, where paint pools, how edges dissolve. That authenticity to physical mediums creates something that feels handcrafted even when AI generated it.

4. Get your hands dirty with post-processing

Raw AI output is raw material, not a finished product. Jake calls his SVG conversion process “a bit of secret sauce.”

You can always tell when someone takes an image straight from ChatGPT and posts it. The artifacts, the color inconsistencies, the subtle wrongness that trained eyes catch immediately. Post-processing removes those tells.

Mia advises learning “just a little bit of Figma” to add what she calls a “human mark”—something small but consistent that represents you. Daria frames it as the difference between “lazy and curated.” She doesn’t hide that her images are AI-generated (she keeps watermarks), but she treats every output as a design element requiring the same scrutiny as any other creative tool.

The goal isn’t disguising AI. It’s applying the judgment that separates acceptable output from distinctive work.

5. Develop systematic prompting and “signatures”

Starting from scratch every time guarantees inconsistency. Anna relies on just two core prompts she’s refined over time—a primary and a secondary that capture her specific aesthetic.

Caitlin McColl maintains consistency through a named character, Elara, with a prompt script that defines her subject and colors while only changing the scenes. The character stays constant; the story varies.

This is about building systems, not crafting one-off prompts. Your “prompt signature” should encode your brand’s visual DNA so thoroughly that even when concepts change completely, the output remains unmistakably yours. Think of it like a style guide, but for AI—a set of constraints and preferences that travel with every generation.

The more systematized your prompting, the more consistent your output, and the less time you waste in the regenerate-regenerate-regenerate loop.

6. Build character and asset reference pools

If you use consistent characters in your visuals, you need to lock that down.

Jenny learned this the hard way. She accidentally used a default ChatGPT model instead of 4o, and suddenly her brand character had short hair, glasses, tanned skin, rnone of which matched her established look.

Without reference pools, your character drifts with every generation and every model update. Hair length changes. Skin tone shifts. Proportions wander. The continuity that makes a brand recognizable disappears.

Dallas Payne offers an alternative: design around AI’s limitations rather than fighting them. Her “faceless navigator” approach sidesteps the facial consistency problem entirely, creating universality while maintaining visual coherence. Karen Spinner advises editing one aspect at a time rather than overwhelming the model with multiple reference images that compete for attention.

Either way, you need anchoring material, whether that’s a detailed reference pool or a character design that avoids AI’s weakest capabilities. The point is intentional architecture, not hoping the AI remembers what you want.

7. Prioritize tool loyalty over novelty

Every AI model has default behaviors: aesthetic tendencies, interpretation quirks, strengths, and blind spots.

Mia advises becoming “a student of the AI visual medium,” learning to spot those patterns so you can intentionally break them. That knowledge takes time to develop. When you switch tools constantly, you’re perpetually starting over, never building the fluency that lets you push a tool beyond its defaults.

Jenny warns that newer models often “improve” in ways that break your specific constraints—adding unwanted details, making assumptions, optimizing for what the model thinks looks good rather than what your brand requires. Daria notes how easy it is to get distracted by how cool a new AI style looks, letting the tool dictate style rather than brand strategy.

Brands are built through repetition, not novelty. Once you find a tool that delivers consistency, learn its quirks deeply rather than chasing the next shiny release.

The outcome is Unpromptability

The difference between a stock-photo-factory brand and a studio-quality brand isn’t the AI you use. It’s the posture you bring.

Nick Quick said it best: “The fingerprints matter. The factory they came from doesn’t.” When you push until the output matches your vision—or make it yourself when the AI won’t cooperate—the tool becomes invisible. What remains is distinctly, recognizably you.

That’s Unpromptability in visual form.

Anyone with a subscription can generate images. Anyone can type “professional brand aesthetic” and accept what the algorithm hands them. But the creator who shows up with Human Direction, builds a Visual System, and infuses every piece with Artistic Soul—that creator can’t be copied. Not by AI. Not by competitors. Not by anyone with better tools and more budget.

Your visual brand is part of your moat. Treat it like one.

PS: If you want help building AI systems that help your business become defensible — that’s what I do. I bring imagination (connecting automation to moats, which others don’t) and implementation (translating strategy into working systems).

In Q1 2026, I’m opening up only 3 slots for founders who want to become Unpromptable. If you’re interested, join the waitlist. There’s only one spot left.