While AI changes how marketing gets done, Jeremy MacDonald argued, it is quietly changing the job itself. Teams are shifting from doing the work to engineering the structure that does it. His name for the new role is the marketing engineer: someone who builds a system that does the work and isn’t bound by the usual limits of time and effort.
This is the final post in my series on his BrightonSEO talk. If you’re just arriving, start with what a moon rocket teaches AI teams.
What a marketing engineer builds
His keyword research demo is a small example: a chain of generation steps with an eval at every stage (I walked through it in build your first agentic eval). His examples of where else it applies were content briefs and support tickets, and he said some of the systems his company builds are advanced while others are simply skills, depending on the complexity.
The larger example was a tool built for an FMCG insights client. The brief was to work out how to reach younger consumers for a casual dining brand. Previously the team produced and read many reports. Now the tool runs this flow:
Nineteen evals run end to end, and Jeremy said each output is checked about 20 times. On the front end it looks like a normal web page with links so users can click through to the evidence. What the team used to get from their best person in a week now runs at a scale no person could. He said products driven by the tool are due in large fast food chains within six months, a claim that comes from his talk.
The skills the role needs
The pattern in his examples suggests four skills:
- Process design. Writing down the best version of a task, step by step (the “if I had a week” exercise).
- Prompt structure. Using a framework such as RISE so prompts are repeatable.
- Evaluation design. Defining criteria and checking the checker (see who grades the grader?).
- Data sourcing. Knowing which data to add that people can’t normally reach.
Those are senior skills. It’s one more reason I think AEO teams should be senior-only.
The bigger point
Jeremy’s closing argument was that thinking about how to do your old job better with AI is the wrong goal. The goal is how to do a job that wasn’t possible before. In his words, that’s what AI will mean for your career: not just doing work cheaply and quickly, but doing revolutionary work.
For teams starting out
- Pick one process you run often and write it as steps.
- Add an eval after the most error-prone step.
- Run a feedback loop for a month and track the score trend.
- Share the system, not just the output, so others can reuse and improve it.
Want help engineering your first AI marketing system? See our services or get in touch.
Leave a Reply