Point of view

The CMO who builds

What changes when the person leading growth also builds the systems that run it.

By Julien de Waal  ·   ·  4 min read

“AI CMO” usually means software that replaces a marketing leader. This is the other thing: a marketing leader who builds the AI.

I spent years running growth the traditional way. At SwissBorg I led Growth for the Web Trading platform: a CHF 1M+ budget, 12 direct reports and about 20 cross-functional partners across product, engineering, data, legal and compliance. Since then I have built the agent systems I wished that team had. This is what doing both has taught me.

My bet

The best marketing teams of the next few years will be small groups of strategists directing agent systems. Not teams replaced by software, and not teams that use a chatbot for first drafts. Teams whose operating model is built around agents doing the repeatable work, with people owning the decisions that matter: what to build, who it is for, what to stop.

I cannot make that argument credibly from a slide. So I build the systems myself, and I lead teams the way I always have. Each side keeps the other honest.

What I did inside a company

At SwissBorg I used AI where volume was the bottleneck. We shipped 300+ AI-powered SEO pages a quarter, built AI workflows for token launch pages and content, and moved non-branded organic from 5% to 35% of traffic, reaching 500K+ quarterly organic visits. The full case study is here.

The lesson was not about tools. The tool was the easy part. The hard part was the operating model around it: who briefs, who reviews, how pages ship and how you measure them.

What I built on my own

None of these has a marketing team. Each has a system, and me.

What agents do well, and what I keep human

Agents are good at work that is repeatable, well specified and high volume: pulling data, drafting, formatting, adapting one idea to five platforms, assembling files, publishing on a schedule. They are poor at deciding what is worth making.

Taste, positioning, the call on what to ship and what to kill, and accountability for spend stay with a person. That is why, in my own systems, some parts run autonomously and others keep me in the loop. It is also why I distrust anyone who says the whole thing runs itself.

The failure I watch for is quiet: output that looks fine and says nothing. Volume is easy to generate and hard to make matter. So I measure outcomes, not output.

How I would build a marketing team around this

And track cost per output. The Sonscape pipeline produces a finished video for under $100, against 40+ hours of manual production. If you cannot state the cost per output of your AI workflow, you do not have a system yet. You have a habit.

Five questions to ask a CMO who says they are AI-native

  1. Which part of your growth stack runs without you? Show me.
  2. What does it cost per output?
  3. Where does it fail, and who catches it?
  4. What did you stop doing because of it?
  5. Could you rebuild it if the vendor changed the model?

Work with me

If you want a leader who builds, or a system built, write to me. I take on CMO, VP Marketing and Head of Growth roles, fractional CMO work, and AI agent system builds.

Talk about a role  ·  Scope a build