AI changed the cost of execution, not the source of judgment
For most of marketing's history, execution was the constraint. Producing the campaign, writing the content, building the program, running the analysis: all of it took time, people, and money. The scarce resource was the capacity to do the work, and a great deal of marketing leadership was really about managing that scarcity, deciding what the team could realistically produce and in what order.
AI has dramatically removed that constraint. Content, research, campaign variations, experimentation, analysis, outbound: the things that used to require headcount, agencies, and weeks now take a fraction of the time and cost. This is the most significant leverage marketing has had in a generation. Any company treating it as optional is making a mistake it will pay for.
But here is what the moment is obscuring. AI changed what execution costs. It did not change the source of judgment. And judgment, not execution, is now the thing in short supply.
What got cheap, and what didn't
Execution is the doing. Judgment is deciding what is worth doing. AI is extraordinary at execution. It can support judgment, pressure-test options, and surface patterns, but it cannot own the judgment itself, and the gap between assistance and ownership is the whole story.
AI can write a hundred variations of a campaign. It cannot tell you whether that campaign should exist, or what you will stop doing to make room for it. AI can produce a quarter's worth of content in an afternoon. It cannot tell you that the content was never the constraint. It can analyze your funnel faster than any human. It cannot decide which problem to solve first when the budget will not cover all of them.
It can even connect your tools, wiring the systems together faster than a specialist team once could. What it cannot do is connect your activity to your revenue, because that connection is not a technical one. It is a set of decisions about what counts, what matters, and what the business actually needs.
Those are judgment decisions. They depend on knowing the ICP, the commercial priorities, the revenue math, and the specific situation of the business. They are exactly the decisions a strong operating model exists to govern. AI does not supply them. It executes against them, and it executes against whatever it is given, good model or bad.
That is why output is no longer the differentiator. When everyone can produce more, faster, producing more stops being an advantage. The advantage moves to the quality of the judgment directing the output: whether the activity is pointed at the right buyer, the right segment, the right motion, the right number. Leverage applied to a system that connects activity to revenue compounds. Leverage applied to a system that doesn't just produces noise at scale. (I have written separately on how a gapped operating model gets worse, not better, with AI; this is the principle underneath it.)
AI-enabled, human-led
This is what I mean by AI-enabled, human-led, and it is not a slogan about keeping humans in the loop for comfort. It is a statement about who holds the judgment AI cannot supply.
The human leads because the human owns the model: the strategy, the ICP, the priorities, the tradeoffs, the call on what to pursue and what to decline. That is the judgment layer, and it is the part AI cannot author. AI is enabled against that model to do what it is genuinely better at: producing at a speed and scale that used to require a team three times the size. Get the order right and AI is a multiplier on good judgment. Get the order backward, lead with the tooling and hope judgment follows, and you automate a function that was never pointed in the right direction.
The companies getting real leverage from AI right now are not the ones with the most sophisticated stack. They are the ones with the clearest model underneath it. They know who they are targeting, how revenue is created, what matters this quarter, and where accountability sits. AI makes that system faster. It does not create the system, and it cannot.
Where this leaves the work
None of this is an argument for caution about AI. It is the opposite. Because execution is no longer the bottleneck, the return on getting the operating model right has never been higher. The model is what turns cheap execution into compounding advantage instead of expensive noise. That is the work I do: building and leading the operating layer that connects strategy to execution, with AI as one of the most powerful tools inside it, not a replacement for the judgment that directs it.
I hold this position because I have built on it. Acton Hunter itself was built with this principle: AI as leverage, human judgment as the operating authority. AI helped accelerate the frameworks, systems, and tools of the practice, but every consequential decision, the positioning, the service architecture, the commercial model, and the standard for what good looks like, is led by human judgment. The principle is not theoretical to me. It is how the business exists. And it is the same principle I bring to the companies I work with: AI makes a good operating model faster, and it makes the absence of one impossible to hide.
If your team is producing more with AI but pipeline is not growing, the gap is not the tooling. It is the operating model the tooling is landing in. A GTM Revenue Audit surfaces where that model is breaking down.
See how Diagnose works →