It’s never the question I expect. Not “will AI replace me” in some abstract thinkpiece way. The version I actually hear is quieter and a lot more personal.
It usually comes about forty minutes in. I’ve walked an agency team through what’s possible when AI gets integrated across the tools they already use (Slack, Asana, Drive, ad platforms, analytics) and the room has watched a picture come together that would’ve taken four people a week to assemble by hand. The room is quiet for a beat. And then somebody laughs a little, looks around, and says it.
“Do I still have a job?”
There’s always a kernel of truth in that joke. I’ve had some version of that moment in a lot of agency conversations over the last year, and the laugh is always the same. Uncomfortable. Knowing. A little hopeful that I’ll reassure them and a little aware that no reassurance is going to land unless it’s actually true.
Most of what’s being written about AI in agencies right now is missing the thing I think is actually true. So I want to say it plainly.
The Fear Underneath the Fear
The surface fear is obvious. AI does the deck. AI writes the brief. AI builds the first round of concepts. If the machine can do the work, what’s the person for?
But when I sit with agency folks long enough, a different fear surfaces. The real fear is that AI is going to take the part of their job they actually love. The research rabbit hole. The concepting session at 10pm when it finally clicks. The Slack thread with a strategist you trust where you work out what the client actually needs. The parts that reminded them why they got into this work.
I get it. If you’ve watched the first wave of AI-generated content, there’s a flatness to a lot of it. A sameness. You can feel it in the body of a blog post three sentences in. You can feel it in the pitch deck a competitor just put out. And if you’re being honest, you’ve probably seen work from inside your own agency that had a little too much machine in it and not quite enough person.
So the fear is legitimate. I’m not going to tell you it isn’t.
What I’ve Actually Been Watching Happen
I’ve been in enough of these rooms to say something with confidence. The agencies producing flat, forgettable AI work have one thing in common. They stopped thinking. They treated the tools like a vending machine. Brief in, deck out. Ship it.
The agencies wowing clients right now are being wildly more creative and more strategic than agencies doing this work a year ago. Making AI actually behave inside an agency, the way you’d want it to, takes a kind of creative and strategic thinking most of us have never had to do before.
When we built our Client Knowledge Agent at Newfangled, the hard part wasn’t the AI. The hard part was everything else. Which signals from Slack should talk to which data in Analytics, and when? What does the system surface proactively versus wait to be asked about? What does it refuse to do? How does it sound when it talks back to a strategist? What does “a good recommendation” even look like for a client services team? How do we make sure it catches the mismatch between what the client said in a Slack thread last Tuesday and what their services page actually says today?
Every one of those is a creative decision. Every one of those is a strategic decision. There isn’t a model on the market that makes those calls for you. There’s no tool you can buy that skips this part. Somebody sat down and decided what good looked like, and then built a system that behaves that way.
That’s the work. And it’s the most creative work I’ve done in years.
The Creative Work Got Bigger
If you’ve spent your career developing taste and judgment, an AI-first agency needs you more than a traditional one ever did. Strategists who can smell when a brief is wrong. Creative directors who can tell the difference between a concept that almost works and one that actually does. Account leads who read what a client isn’t saying. You are the layer the system cannot replace and cannot be built without.
Because when the raw output is a commodity, everything around the output becomes the differentiator.
What prompt you wrote. What you let the system do and what you insisted a human do. What you chose to surface to the client and what you chose to refine one more round. What you decided the brand voice actually sounds like, written down in a way a model can follow. What patterns you noticed across clients that the system wouldn’t have connected on its own. What you did with the two hours you got back.
Those are creative acts. They’ve always been creative acts. We just used to bury them under so many billable hours of execution that it was hard to see where the real value was.
So, Do You Still Have a Job?
Yes. Probably. But I’m not going to pretend it’s the same job.
If the part you love is the craft in its narrowest sense (the writing itself, the building of the thing), some of that is going to shift. The tools are getting good at the outputs and they aren’t going to get worse. That’s real.
But if the part you love is the feeling of making something that surprises a client. Something that fits. Something that works in a way nobody could have predicted from the brief. That feeling is about to be more available than it’s ever been. The taste and creative system design that produce it are exactly what an AI-first agency needs more of.
The people I’ve watched thrive over the last eighteen months share one thing. They stayed curious. They treated every new capability as a chance to design something smarter. Tool fluency didn’t predict who would do well. Curiosity did. Fear shuts that disposition down. Curiosity opens it up.
The agencies I’m most optimistic about share a posture. They’re willing to do the harder creative work on top of the tools, instead of hoping the tools do the creative work for them. Willingness is the variable. Always has been.
That’s the job. It was always the job. We just have better instruments now.