$5,500 And A Month Of My Life For A Website. Last Week I Built Four In An Hour.

Years ago, I hired a web designer.

Three grand. For mockups. Not code. Not copy. Just the design.

Ten revisions in, the guy still didn't get my vision. Revision 10 was the one we went with. Not because it was good. Because it sucked the least.

(I know. Sad.)

Took that mockup over to a WordPress developer. Paid him another $2,500 to actually build the thing.

Then I still had to sit down and write every word of copy myself.

Almost a month of my life before I had a working website.

$5,500 total.

For a website.

Ok, so last week I built four of them. All in under an hour. One in under 15 minutes.

Full stack. Designed. Coded. Done.

I did it by talking into my microphone.

Same quality as what I used to pay five grand for. Honestly... better.

And Then GPT Image 2.0 Dropped

If you've been online this week, you've seen the noise.

OpenAI released GPT Image 2.0 a few days ago and it's genuinely nuts. Near-perfect text rendering. Pixel-perfect UI mockups. The weird "AI look" that used to give everything away is basically gone.

That $3,000 mockup I paid for back in the day?

I can knock that out with a single prompt now.

But here's where most people are about to miss the real story.

GPT Image 2.0 by itself makes really good generic graphics.

Better generic. Still generic.

It's not the model that matters.

It's what you feed it.

The File That Does The Heavy Lifting

Last week I wrote about the Brand DNA app I built. Five minutes of guided questions. Pick your colors. Pick your fonts. Answer a handful of things about your voice and your ideal customer.

What you get out is a single markdown file. Your complete visual brand system in a format any AI can read.

Ok so here's the part I didn't fully explain last week...

That file works with everything.

Not just GPT Image 2.0. Everything.

I've been feeding it to ChatGPT for GPT Image 2.0 graphics. They look like I hired an agency.

Last week I handed the same file to Claude and had it spit out a full carousel. (Yeah, Claude does images now too. Most people don't know that yet.)

I'm generating graphics in Manus with the same file. Manus uses Nano Banana, which GPT Image 2.0 just absolutely smoked this week, by the way. Race to the top imaging model isn't slowing down either. New models every couple weeks. Capabilities moving constantly.

Doesn't matter.

Same file works in all of them. Different tools. Same brand. Every time.

Chasing Tools Is A Losing Game

I want you to actually sit with this for a second.

The tools are gonna keep changing. Fast.

GPT Image 2.0 is hot this week. Six weeks from now, Google will drop something that smokes it. Then Anthropic. Then somebody we haven't heard of yet.

That's just how this is gonna go from here. Forever.

You could spend the next two years chasing whatever the new hotness is. Learning every new model. Rewriting every new prompt. Watching every YouTube "OMG THIS CHANGES EVERYTHING" video.

Or you could build one asset, once, that works with every tool that comes next.

Your colors. Your typography. Your visual style. Your tone. What you want your perfect customer to feel when they see your stuff.

Build it once. Use it forever. Load it into whatever's new next month.

(Quick side note... I handed this file to my AI agent team and they're running it across Claude, GPT Image 2.0, and a handful of other tools to crank out branded visuals at scale. That's way further down the rabbit hole than most people need right now. Start with the file. Worry about the agent stuff later.)

Now Here's Where It Gets Real

Colors and fonts are the easy stuff. The visible stuff.

But there’s something far more important than that…

The emotional association.

What do you want your ideal customer to feel when they see your stuff?

Confidence. Trust. Rebellion. Calm. Urgency. Authority. Warmth. Whatever it is.

That feeling... that's what a brand actually is.

Not your logo. Not your color palette. Not your font pairing.

It's the subconscious thing that happens in someone's brain the moment they see your content scroll past. That half-second before they've read a single word.

That feeling is what turns cold traffic into warm leads. It's what makes a premium price feel reasonable. It's what makes a generic competitor feel cheap by comparison.

Your brand is the thing people feel about you when you're not in the room.

And if AI doesn't understand that feeling, it's not gonna matter how powerful the model is.

It'll produce generic stuff that looks like everyone else's generic stuff.

More capable AI + no brand context = fancier garbage.

Same brand context + multiple tools = a design team that never sleeps.What the Brand DNA App Actually Does

Ok So Here's What To Do

Five minutes. Guided questions. One file.

Drop it into ChatGPT for GPT Image 2.0 graphics.

Drop it into Claude for carousels.

Drop it into Manus. Drop it into Gemini. Drop it into whatever drops next month that everyone loses their minds over.

Generate stuff that actually looks like you. Not like the same AI slop everyone else is posting.

Stop paying $5,500 for mockups. Stop staring at generic AI graphics that don't feel like your brand.

Try The Brand DNA Extractor

It's free. Takes five minutes.

One More Thing

I'm thinking about doing a full video walkthrough on this.

Showing exactly how to use the Brand DNA file across GPT Image 2.0, Claude, and a couple other tools. Including how to take a static graphic and animate it into a full motion video for social (yeah, you can do that now too).

If that sounds useful, just hit reply and say "video."

If enough people do, I'll shoot it.

Until next time,

—Tim Erway

P.S. If you know somebody about to cut a $5,500 check to a designer for something AI can knock out this weekend... send them this. They'll either thank you or pretend they never saw it. Either way, you did your civic duty.

P.P.S. The tools are gonna keep changing. GPT Image 2.0 is the hotness this week. Something else is gonna be the hotness next month. The Brand DNA file works with all of them. That's the part worth building.