During this job search, I've been reading a lot about what AI can do for marketers. This tool automates your research. That tool writes your first draft. Another one builds your competitive matrix while you sleep.
As a product marketer with more gray hair than black, my first reaction was pretty human: is AI coming for my job?
It's not. But that fear is worth sitting with for a minute, because it tells you something about where the real value actually lives.
Here's what I've figured out, both from using these tools heavily over the past year and from watching how other senior PMMs are starting to work: AI is genuinely good at scale tasks. Research, synthesis, first drafts, formatting, competitive monitoring, and content variants. Things that used to eat half your week.
What it cannot do is the actual job.
Product marketing at its best is pattern recognition applied under pressure. It's sitting across from a founder who is too close to the product and finding the three words that make a skeptical CISO lean forward. It's knowing which competitive narrative will land with a practitioner audience versus a board. It's reading a room at a conference and adjusting the message in real time because something isn't connecting.
That judgment doesn't come from a prompt. It comes from years of being wrong in front of real buyers and learning from it.
So here's the framework I've landed on. AI plays one of three roles depending on the task.
The Collaborator
Recently, I was building a messaging framework for a startup from scratch. No formalized messaging existed, but the company had a clear sense of how it wanted to be perceived. I started the way I always have, writing a rough narrative covering the problem, why current approaches fell short, and what value the product actually delivered. Raw thoughts, not polished copy.
Before I went any further, I asked my AI tool to go find out what was publicly available about the company. The results were immediately useful. It surfaced references to product features and names that no longer existed, old artifacts still floating around the internet. I checked the website, sure enough, some of it was still there. We cleaned it up before the messaging work went any further.
From there, it became a back-and-forth. I shared my raw messaging and asked where it diverged from what was already out in the market. Then I pulled in three competitors with similar offerings and asked how they were positioning. Then I pushed on the one thing I knew we did that nobody else did, and asked the AI to challenge me. It found some overlap I hadn't seen. More data.
All of that brought me to a real insight: every competitor was following industry best practices that, honestly, made things more complicated than they needed to be. We had a more direct path. My brain went to the Matrix. Red pill, see the truth, reject the conventional approach. I spent a week iterating on that premise with my AI tool, hundreds of variations, and landed on something that felt genuinely sharp.
Then I asked one more question: What would a CISO who has been burned by overhyped startup claims think of this message?
The answer stopped me cold. Not everyone wants to rage against the machine. Some just want things to work better without blowing up what they already have. That was the real pivot. I dialed back the revolutionary framing and replaced it with something simpler: you don't have to follow best practices, and you don't have to change everything at once. Try our approach. See what happens.
Another week of iterations, and the framework came together.
Could I have done all of that without AI? Yes. I did it for years. But it would have taken months, not weeks. And when that strategic pivot happened mid-process, updating every artifact in the old world would have been a nightmare. With AI as my collaborator, it was an afternoon. I didn't write every word of that framework. But I had final say on all of it, and that distinction matters.
The Intern
Not every task needs a collaborator. Sometimes I just need scale work done fast. Competitive research across ten vendors. A summary of analyst coverage. Five subject line variants for an email. A first-pass data sheet based on a spec I handed it.
This is where AI functions like a smart intern. It can move quickly, cover a lot of ground, and produce something usable. But like any intern, the output needs a senior set of eyes before it goes anywhere. The research might miss context. The draft might be technically accurate, but tonally wrong for the audience. The summary might hit the facts without understanding what actually matters.
The value isn't that the work is done. It's that the starting point is miles ahead of a blank page, and I can apply judgment to something concrete rather than build from nothing.
The Editor
The third role is the one I use most often on shorter work. I write something, then I hand it to AI and ask for an honest read. Is this clear? Is it too long? Does it sound like a human wrote it, or does it sound like a press release? Where is it losing the reader?
This is harder to do well with a human editor because it requires someone willing to be direct, available on short notice, and fast. AI checks all three boxes. I don't always take the feedback, but having something push back on a piece of writing before it goes out has saved me from more than a few regrets.
What This Adds Up To
When you roll it together, you basically have a team working under your direction. A research associate, a drafting resource, and an editor on call. The senior judgment, the market instincts, the message architecture — that's still yours. AI removes the ceiling on what you can produce without removing the thing that makes the output worth producing.
The PMMs who are thriving right now aren't the ones ignoring these tools or the ones outsourcing their thinking to them. They're the ones who figured out which parts of the job require a human and protected those fiercely.
That's the calculation worth making.