Marketing is changing fast. There’s seems to be a brand-new type of marketing tech in town – and they don’t just enable marketers, they do the job for them too, or so suggests a recent feature published by The Marketing Centre.
They launch campaigns, move budgets, change CRMs, post content and keep the engine running with barely any human direction.
That sounds efficient, right? But here’s the rub: if these agents and intermediaries assume all the doing, who’s left to do any thinking – to strategise, make voice, craft judgment calls?
The article poses this question outright: “If everybody is automating different parts of their job away, who will be left to do the tasks that can’t be automated?”
Many organizations, marketers are used to working with more basic AI tools: chat-based custom GPTs for brainstorming and research, or copy.
There were actual prompts, and humans to follow through. The turn now is toward agents who act, instead of waiting.
They plug into workflows, connect to CRMs, ad platforms, analytics system, basically manage full campaigns.
I’ll confess – part of me is thrilled at the potential, the prospect of releasing human marketers from needlessly repetitive grind work to allow them to concentrate on bigger creative and strategic challenges.
But another part of me is cautious because the autonomy carries real risks. And when an AI agent is posting that clip, making a budget adjustment or determining what case to fight, the risk of something going off-brand and leaking sensitive data, or just misfiring in some way, increases significantly. The article references this: “with autonomy comes risk.”
And there’s more than just internal risk here. A recent study commissioned by Capgemini published Globally in July 2025 found almost 70% of marketing leaders believe that autonomous or multi-agent AI could be transformational – but just a tiny majority claim they have experienced strong marketing effectiveness from their use.
So here’s my conviction, based on what they tell me: the winners in this next phase won’t be those who automate fastest.
It will be those who govern best, who build structures that protect brand integrity, ethical compliance, data security – and the human spark.
The doing will be mostly by machines; the thinking and tone and strategy – that’s still humans.
A few questions I’d ask right now if you’re a marketing lead:
What aspects of our process should not be automated because it needs that human touch, or brand nuance?
How are we instructing agents who manipulate live systems?
Is our workforce increasingly organising AI workflows, as opposed to physically performing actions themselves?
Do we have any kind of governance construct – permissioning schema, audit trails, a brand review – before the agent has live access?
Because, let’s face it, delegation of the “doing” to machines can be great – but if we delegate the “thinking” with no plan in place, that’s dangerous rocks-ahead territory.
Ultimately: agents will take over most of the doing, but humans will always own voice, values, story, creativity.
And if lose that, we lose what makes marketing human. And, to be honest, that’s what I’m afraid we’re going to automate away.
