back to blog

brand

Why AI Can't Replicate Your Brand (And Why That Matters)

You can't describe the visual difference between Nike and Adidas in a prompt. And when you try, people notice. Here's why 'close enough' is actually worse than starting from scratch.

Mike Cornelius · January 26, 2026 · 10 min read

Last week, I asked an AI to "create a presentation in Nike's brand style." What I got back was... close. The swoosh was there. The colors were roughly right. But something was off. It looked like Nike the way a wax figure looks like a celebrity - technically accurate, deeply unsettling.

This is the dirty secret of generative AI and brand work: close enough is actually worse than completely different. When something looks almost right but not quite, it triggers what psychologists call the "uncanny valley" - that visceral discomfort we feel when something mimics humanity but doesn't quite nail it.

And when that uncanny valley applies to your brand, the consequences aren't just aesthetic. They're relational.

The CSS Problem: Why Brands Can't Be Prompted

Let's get technical for a moment. When you build a website or presentation, the visual identity lives in something like a CSS file - a set of rules that define colors, spacing, typography, shadows, gradients, and hundreds of micro-decisions that collectively create a "feel."

Both brands use black and white. Both are athletic. Both are premium. But the differences are in the details:

  • Letter spacing: Nike pulls letters together (-0.03em) for tension. Adidas spreads them apart (0.02em) for openness.
  • Shadows: Nike uses dramatic, high-contrast shadows. Adidas uses softer, more diffuse ones.
  • Animation: Nike is snappy and aggressive (0.2s). Adidas is smooth and athletic (0.35s).
  • Corners: Nike is razor-sharp. Adidas has subtle rounding.

Now try to describe that difference in a prompt. Go ahead. I'll wait.

The Prompt Problem

"Make it look like Nike but for our brand" gives you black backgrounds and bold fonts. It doesn't give you the specific cubic-bezier timing that makes a hover state feel athletic. It doesn't give you the exact shadow depth that makes a product shot feel premium. The difference between Nike and Adidas isn't in the words - it's in the 200 micro-decisions that no prompt can capture.

The Uncanny Valley of Brand

The uncanny valley was originally about robots and CGI humans. When something looks 95% human, it's more disturbing than something that looks 50% human. Our brains are wired to detect subtle wrongness.

The same principle applies to brands. When your AI-generated deck is 95% on-brand, it's worse than being completely off-brand. Here's why:

95% On-Brand (Uncanny Valley)

  • Feels like you tried and failed
  • Suggests carelessness or laziness
  • Makes people question your attention to detail
  • Creates cognitive dissonance - "something's wrong"
  • Worse: they can't articulate what's wrong

100% On-Brand (Invisible)

  • Feels professional and intentional
  • Suggests care and investment
  • Builds trust through consistency
  • Creates no cognitive load - it just works
  • The brand becomes invisible (which is the goal)

The Trust Problem

Here's where it gets real: when you send a client or internal stakeholder a deck that's obviously AI-generated, you're not just making a design mistake. You're making a relationship mistake.

Think about what that deck communicates:

"I valued our relationship so little that I spent 30 seconds typing a prompt instead of actually thinking about your business."

We've all seen the LinkedIn posts. The ones where someone proudly announces "I USED AI TO MAKE THIS" in the comments, as if that's an accomplishment. As if the tool is the story.

But here's what your clients are actually thinking:

  • "If they're using AI shortcuts for the deck, what else are they shortcutting?"
  • "This doesn't look like us at all."
  • "Did they even look at our brand guidelines?"
  • "I'm paying for expertise, not prompts."

The trust erosion is real. And it's happening faster than most people realize because AI-generated content is becoming easier to spot, not harder.

How Decky Approaches This Differently

When we built Decky, we started with a simple observation: your best brand asset is your existing brand asset.

You've already made the 200 micro-decisions. They're already encoded in your existing decks, templates, and materials. The Nike deck already has the right letter-spacing, the right shadows, the right animation timing. Why would we try to recreate that from a prompt when we can just... use it?

Decky's Approach: Reuse, Don't Recreate

Instead of asking AI to generate your brand from scratch, Decky works with your existing materials:

  • Your slides become the source of truth. We analyze your existing presentations to understand exactly how your brand looks - not approximately, exactly.
  • We intelligently decide what to update. New data goes in. New logos get swapped. Dates get updated. But the layout, the spacing, the feel? That stays exactly as your design team intended.
  • AI is used where AI belongs. Interpreting data, generating charts, pulling the right metrics - places where you don't care about AI because the output is functional, not aesthetic.

Where AI Belongs (And Where It Doesn't)

We're not anti-AI. We're anti-AI-in-the-wrong-places. Here's how we think about it:

AI Should Handle:

  • Interpreting raw data into insights
  • Generating charts and visualizations from data
  • Swapping out logos based on audience
  • Updating dates and timelines
  • Pulling content from your team's working documents
  • Summarizing meeting notes into bullet points

AI Should NOT Handle:

  • Deciding your brand colors
  • Choosing typography and spacing
  • Creating layouts from scratch
  • Generating "on-brand" imagery
  • Determining visual hierarchy
  • Making aesthetic judgments

The distinction is simple: AI is excellent at functional tasks and terrible at aesthetic ones. Use it where its weaknesses don't matter.

The Result: Your Deck Actually Looks Like Your Deck

When you use Decky, the Nike deck looks like Nike's deck. Not because we're good at prompting. Not because we have a magic algorithm. But because it literally is Nike's deck, just updated with the right information.

The Q4 numbers are current. The customer logos are swapped. The agenda reflects this week's meeting. But the shadows are still exactly 40px with 15% opacity. The letter-spacing is still -0.03em. The animation timing is still that specific cubic-bezier curve that makes it feel like Nike.

Your design team spent months getting those details right. Why would you throw that away for a prompt?

The Bottom Line

Generative AI is remarkable technology. But it's not brand technology. The visual language that makes Nike feel like Nike - or makes your company feel like your company - can't be captured in a prompt. It can only be preserved.

Stop trying to describe your brand to AI. Start reusing what you've already built.

Written by Mike Cornelius, Co-founder of Decky. We build tools that keep your brand intact while automating the tedious parts of presentation creation.