The Bottleneck Isn't The Tools
A few events this week, and some reading in between, got me thinking about where marketing actually is with AI right now. Not where the vendors say it is. Not the conference version. The real version, from conversations with somewhere between two and three hundred CMOs and technologists over the past few months.
Here's what I'm seeing. Marketing has to keep pace across more segments, channels, and markets than ever before, at a point in time where the noise has never been louder. The volume of messaging, positioning, and content out there is at an all time high. And it's not slowing down.
Real progress has been made with AI in copy and personalisation. That's genuine. But the gains are often localised, or worse, viewed with suspicion by the teams who didn't choose the tools and weren't brought along in the thinking. Faster outputs haven't produced faster execution, because the bottleneck isn't the tools. It's the operating model. And when faster outputs are lazy or ill-advised, they reinforce exactly the suspicion that slows adoption down. It becomes a cycle that's hard to break.
On the tools themselves: businesses have never had so much choice. That's good for pricing and good for innovation. But it's awful for architecture strategy, for efficiency, and often for spend, especially in large enterprises where tools get acquired without centralised decision making or any real centre of expertise. The tech stack becomes a sprawl, and no one quite owns it.
Meanwhile, sequential processes and siloed systems can't move at the speed the business now demands, particularly when teams can't rehire or secure new headcount. And Founders and CEOs now believe entire marketing organisations can be replaced by AI. It's a perfect storm, and most marketing leaders are trying to navigate it without a map.
—————————————————
The organisations thinking ahead are building something different.
They're building what I'd call an agentic marketing organisation, where autonomous workflows run on a shared foundation of intelligence. At the core sits something I think of as the brand code: a machine-readable knowledge base that captures strategy, product, customer insight, and business rules in a format that both people and agents can actually act on.
It's the always-on onboarding for the whole system. And it sharpens with every campaign, as long as someone keeps it updated.
Building on that foundation, marketing effort resolves into five connected workstreams. Agents handle execution and coordination. Marketers set intent and judgement.
Intelligence, ideation, signal. Agents synthesise market signals, competitive intel, audience behaviour, and performance data into prioritised opportunities and briefs. Marketers evaluate them and set strategic direction.
Content creation. Agents generate on-brief content across formats and segments, drawing from the brand code. Marketers shape creative intent and push the work where it needs to go.
Research and testing. Agents design and run experiments continuously, using real or synthetic audiences. Marketers define the learning agenda.
Distribution. Agents adapt, schedule, and deploy across channels, markets, and segments. Marketers own channel strategy and partnerships.
Performance and reporting. Agents monitor in real time, flag anomalies, and feed learnings back into the system. Marketers interpret the patterns and guide how the system evolves.
—————————————————
What does this mean for marketers?
The role changes fundamentally. Most marketers built their careers by doing the work: writing the copy, owning the asset, running the handoff. In an agentic setup, that control shifts from execution to direction. Marketers become directors of work. They set intent, evaluate outputs, and make judgement calls across human and agent teams.
And it becomes ever more important to be selective with new tools, of which hundreds launch every month.
The hardest part isn't learning new skills, though. It's two things.
First, letting go of the instinct to step in and produce. The marketers who adapt fastest aren't the most technical. They're the people who can recognise quality in context, know when output is good enough, and translate what they learn into changes the system can carry forward.
Second, having people who understand the marketing fundamentals well enough to step in as the human in the loop, and correct anything the agents have got wrong, while the models continue to mature.
The implications for hiring, development, and management are immediate. Organisations need marketers who think in systems rather than tasks, and managers who review the system rather than the deliverable.
That shift is harder than it sounds. But the teams that make it are going to move at a pace that others simply won't be able to match.