What's left after the machine does the execution

I read a marketing newsletter most mornings, and this week the same argument kept surfacing from different directions. AI is absorbing the execution of marketing, and the open question is what that leaves for the people.

The sharpest version came from a job posting. ClickUp is hiring a solo CMO, paying somewhere between $500,000 and a million dollars, to run its entire marketing function with a team of AI agents. One person, a lot of machines. The framing in the writeup stuck with me. AI compresses the execution, so the value moves to taste and decision-making. You're not paying for the hands anymore. You're paying for the judgment about what the hands should do.

That posting doesn't sit in isolation. The same week, another piece noted that only 36% of Fortune 500 companies still use the CMO title, down from 49% a year ago. The role isn't vanishing so much as being folded into revenue and growth jobs. Read the two together and you get a picture of a function being redrawn. The parts that were about producing and managing output are getting absorbed, and what's being argued over is who owns the decisions.

You can see the same line in the smaller, more practical advice. One piece this week listed the marketing tasks worth automating, and the tell was in the reasoning. The good candidates are the ones that repeat in a predictable pattern and stay easy to review before anything goes public. Meeting recaps, sorting search terms, clustering customer feedback. In other words, automate the work where a human can still glance at the result and catch a problem. The judgment doesn't leave. It moves to the review step.

Here's the part I keep sitting with, because it cuts the other way. As the production gets cheap, the market is starting to price the human parts higher, not lower. Ahrefs published an analysis this week making the case that Google doesn't punish AI content, it punishes bad content, and that heavier AI use tends to track with weaker performance. Substack rolled out a tool that shows readers an estimated human-versus-AI split on any post. Both are signs of the same thing. When anyone can generate competent output, the scarce and valuable things become the judgment behind it and the voice that reads as a person's.

So if you're a marketer watching this and wondering what to invest in, I don't think the answer is learning to prompt faster. The machine will handle more of the execution than it does today, and it will keep getting better at it. The durable skill is harder to name and harder to automate. It's knowing which decision actually matters and being able to defend it, and writing in a way that sounds like you rather than like everyone else's default output.

That's the thread I'll be pulling on with the marketing practitioners I'm talking to this fall. Not what the tools can make, but where the people decide it's worth keeping their hands on the work.

Sources

Back to blog