There's a lot of noise right now about AI in marketing, and most of it sits at one of two extremes: either it's going to replace half the marketing function, or it's a toy that produces generic content nobody asked for. Neither has matched my actual, daily experience using it.
I've spent the last several years building marketing functions from scratch, most recently as Head of Marketing at a B2B technology company, where I was personally asked by the CEO to build the department from the ground up. AI tools like Claude and Notion AI have been part of my daily operating rhythm for a while now, not as an experiment I ran once and wrote a LinkedIn post about, but as tools I actually reach for every day. That distinction matters more than it sounds.
Assisting vs. accelerating
I think the most useful way to talk about AI in marketing isn't "does it work" or "is it good," it's a simpler question: is it assisting you, or is it accelerating you?
Assisting looks like using AI to do a task you'd have done anyway, just a bit faster. Accelerating looks like using AI to get further, faster, on things you genuinely wouldn't have had the bandwidth to do at all otherwise.
In my own work, the clearest example of acceleration was translating deeply technical product value into messaging that actually landed with different audiences: developers, executives, and non-technical buyers alike. That kind of translation work used to eat entire days: understanding the technical detail, figuring out what mattered to each audience, drafting, redrafting. With AI as a genuine research and drafting partner, I could get a strong first pass in a fraction of the time, which meant I could spend the time I saved somewhere that actually needed a human: making sure the tone was right, the claims were accurate, and the story didn't lose its nuance in translation.
AI hasn't replaced my judgment. It's relocated it. I spend less time on the mechanical first draft and more time on the parts of the job that were always the actual value.
Where it doesn't help
I'll be honest about where I don't reach for it. Anything involving genuine relationship building, negotiating a partnership, reading the room in a stakeholder meeting, deciding how to unify a group of executives with very different working styles: none of that gets outsourced. AI can help me prepare for those moments. It can't have them for me.
I also don't treat AI output as finished work. Every piece of content I've used AI to help draft still goes through the same discipline I'd apply to any other first draft: does this say what I actually mean, is it accurate, does it sound like something a real person would say. If I can't answer yes to all three, it doesn't go out.
The unglamorous truth
The organizations that are getting real value out of AI in marketing right now aren't the ones with the flashiest AI-generated content. They're the ones being honest about the boring prerequisite: your data and your processes have to be solid before AI can do anything useful with them. Poor inputs still produce poor outputs, no matter how good the model is. That's not a popular thing to say when everyone wants a quick win, but it's the difference between AI that actually compounds and AI that just produces more noise, faster.
I don't think AI is a novelty anymore, and I also don't think it's magic. It's a tool that's changed where I spend my time, not whether I need to think. For anyone building a marketing function right now, that's probably the more useful conversation to have than whether AI is "good" or "bad."