Hi Reader -

We open-sourced our AI creative strategy agent last week, and I was actually a little nervous about it.

Because a lot of people would be thinking - if not saying this.

"Great. Now you can fire half your creative team, right?"

That's the assumption baked into every "AI agent" launch on LinkedIn right now. You plug it in, watch the headcount line on the org chart get a little shorter, and spend an extra five minutes on the beach.

The honest answer is no. The opposite, actually.

And the reason we put it out there in the first place is that the "AI replaces your creative team" pitch is loud enough already, and the best counterargument we could come up with was shipping the real thing and letting people see what it actually does and doesn't do.

Now, I know what some of you are about to ask, because we've already gotten the question a few times this week.

"Can it just run autonomously and spit out an output?"

Technically yes. It'll give you some output.

We've tried it. You can absolutely turn it loose, walk away, and come back to a sheet full of ad concept ideas(which I'm sure you can connect to your video/image tool du jour via API to get a video or image ad).

--

BUT: is that the ad/concept that's going to unlock performance?

Extremely unlikely.

We know because we ran the experiment. The output is the kind of thing that looks like creative if you squint, and converts like a banner ad from 2014 if you actually run it.

Feel free to try it at home and tell me if you have a different experience.

Soooo. What does it take to use an AI agent well - in a way that actually gets you performance and results?

1. You ask the agent for hooks for a concept.

It surfaces five options. You read all five and kill three of them right away because they're clever and won't convert. The other two you hold side-by-side for a minute, pick one, and then rewrite half of it because the second line is doing too much.

About ninety seconds of work. The agent and the human, both, at every step.

It goes like that across the whole stack.

2. The strategist module surfaces three plausible angles, and you pick the one that matches the actual bet the founder is making on the product.

3. The hook maker generates options, and you kill the cute ones.
4. The concept writer drafts, and you reshape the part that has to ship.
5. The designer suggests first frames, and you direct.
6. The copywriter drafts the script and body, and you edit.
7. The reviewer runs the pre-submission rubric, and you decide what to actually fix versus ignore versus argue with.

The agent surfaces, the human picks.

The agent drafts, the human shapes.

The agent flags, the human decides.

At every single stage.

You get the idea.

--

Here's what the "fire your creative team" pitch keeps getting wrong.

The agent doesn't remove the human from creative strategy. It just removes the busywork.

The blank page. The "what archetypes could I try." The "did I miss anything on the rubric" check. The hour you used to spend staring at a Figma file before you had a single instinct.

What it doesn't remove is any of the parts that actually mattered.

The thinking. The judgment. The taste. The deciding.

Which is why it doesn't shrink your team. It just makes the team you have move about five times faster, on the parts of the job that actually move the business.

Anywayyy, if you spend $50k+/month - and if you'd like to see what it's like to have an agent + human driven workflow unleashed on your ads, hit reply and say 'AGENT' - and we'll set up a time to talk.

Or don't. I'm sure firing your strategist is exactly what you need to do.
​
Later,

Shamanth

P.S. Co-hosting a live session with FunnelFox on June 17 about the click-to-revenue side of growth (the part of the funnel where, in my experience, most of the leak is actually happening). Free, live, real practitioners. June 17, 12 pm EDT / 6 pm CEST.

Register here: https://funnelfox.com/webinar-from-traffic-to-revenue​
​