Jamie Domenici.
Jamie Domenici, CMO, Klaviyo
Meta just started labelling AI-assisted ads. Most platforms will follow. It’s not a question of if but when.
For a while, the big question marketers asked about AI was simple: did you use it? Soon, nobody will care. The question that will matter next is: did it actually make the experience better?
That shift changes more than the toolkit. It changes the job.
For years, the default model for AI in marketing has been human-in-the-loop. It sounds sensible, but it’s built on a flawed idea of what the human is actually there to do — approve things. Sit at the end of the process, waiting for AI to produce something, and sign off on it.
That framing made sense when AI could draft a subject line or resize an image. It makes no sense now that AI can take a single prompt and run an entire campaign off the back of it — audience, creative, send time, channel mix, the lot. If the human’s only job is approval, the human has already been made redundant.
Here’s a better way to think about it: the human is the conductor, and AI is the orchestra.
A conductor doesn’t play every instrument because they don’t need to. Their value is in knowing what the piece is supposed to sound like, understanding the audience in the room, and making the hundred small calls that turn individually correct notes into something that actually moves people. Take the conductor away, and the orchestra can still play… it just won’t mean anything.
That’s the real job of a marketing leader now. Not executing tasks, but setting the strategy AI executes against. Protecting the brand as more content gets produced faster than any human team could review line by line, and understanding the customer well enough to know when something technically works but still feels wrong.
And that last part is where most brands are underestimating what’s coming.
The real risk isn’t that AI lowers the bar for authenticity. It’s the assumption that customers will grade it on a curve. The opposite is true. When your brand is speaking through AI at scale, every interaction is a chance to prove the brand is still actually there behind it. The output has to sound like you specifically, not like marketing in general.
This all brings us back to Meta’s labelling move, and why it shouldn’t spook marketers the way some are treating it.
Transparency isn’t the threat. Skipping the work of understanding your customers is. For everyone else, a label is clarifying. It reframes the real question, for audience and marketer alike: not “was this made by a person,” but “does this deserve my attention.” Brands with a real point of view have nothing to hide — they used AI to execute against customer insight, not replace it. Brands that haven’t should worry less about a label and more about their customers noticing first.
My advice to marketing teams; stop counting the AI tools you’re using. Stop counting the approval steps wrapped around them. Count whether someone is actually conducting. AI running the orchestra doesn’t mean nobody’s on the podium. It means the human job was never about playing every instrument. It was about knowing what the piece is supposed to sound like in the first place.
