Hi Reader -

A lot of people tell me they won’t touch AI creatives with a 10 foot pole. Honestly, fair. I make these things for a living and I still flinch at half the ones I scroll past.

You know the look. That cringe-inducing, uncanny weird aesthetic is real.

I get it.

BUT here’s what everyone misses the second they say they hate AI creatives.

Most people think AI creative means you push a button, you walk away, you come back to a folder of forty ads nobody chose, and somewhere a LinkedIn guy calls it a content engine. A fairy drops dead.

And yes, the bros will say the models have gotten better and more realistic: true.

They can make very realistic 5-10 second clips of AI creators - BUT what the said AI creators need to say? What they need to do? What framing they need to have?

Nope - the robots cant handle that stuff yet.

And that is really why push-button creative just doesn’t convert. Nuanced multi-beat narratives are HARD for robots.

The stuff that actually converts needs a human to be all over it.

There’s a real craft to assembling one of these, and it’s slow and painstaking and frankly kind of annoying. That is exactly why it works. It’s hard, and it’s supposed to be hard.

Let me show you what that means. Before a single frame gets generated, we make a brief, and the brief has more moving parts than the ad does. Here is what a typical brief looks like on our team.

Emotion: shock, then relief. Yes, that’s a field. Somebody fills it in.

Context: one specific setting, picked on purpose.

Audience: who are we targeting?

Hook: someone picks this out of a hundred ‘catchy’/clickbaity options.

Text overlay: the hook line, burned in, so it lands on mute.

The body, beat by beat(example):

1. Every time I tried, I spiked. I was eating the same five things for months.
2. Then I found - I was shocked by what it showed me. (FOOTAGE OF PRODUCT)
3. My ‘no sugar added’ applesauce, 22g of carbs. (TRANSITION TO APPLESAUCE FOOTAGE)
4. My ‘natural’ peanut butter, 12g of added sugar. (TRANSITION TO PEANUT BUTTER FOOTAGE)
5. My ‘whole grain’ tortillas, 28g per serving. (TRANSITION TO TORTILLA FOOTAGE)
6. I finally had a way to figure out how much sugar I was eating. (AI CREATOR FOOTAGE)
7. Now I’m cooking again - and not worrying about what goes into my body. (KITCHEN FOOTAGE)

(that is 7 different pieces of footage that need to be found, curated and stitched together).

Not one of those fields filled itself in. A person picked the setting, wrote the hook, chose the four pairs, went and found the b-roll and linked it to the second, and made the bet on which format gets the budget.

The model didn’t open a tab and go hunting for the right reference clip.

A person did, because a person knows which “looks healthy” actually fools people.

This is what short-circuits the “RIP designers” crowd. We went all in on AI this year, and then we doubled the creative team. Not despite the AI. Because of it.

(I wrote about why here.)

AI let us make ten times the footage, which made the people who can tell good from garbage ten times more valuable. Funny how that works.

If you’re spending $50k+/month and you want AI creative that performs instead of AI creative that makes your brand look like it gave up, reply “REAL” and we’ll talk about running it on your account.

Or don’t. I’m sure the magic button that spits out forty ads nobody chose is one prompt away from cracking your CAC.

Later,

Shamanth