TL;DR In this article I share how I’ve started using AI in my day-to-day work as a product marketing lead, where it actually helps, and more importantly, where it falls flat.
The first time I saw a Slack message suggesting I “leverage AI for competitive intelligence,” I cringed. Not because AI doesn’t belong in competitive intelligence. It does. But because the suggestion treated it like a feature flip rather than a rethinking of how I actually spend my time.
I’m a Certified AI Product Manager and a PMM Lead. That credential isn’t a flex. It just means I’ve spent enough time understanding how AI systems actually work to know when hype outpaces reality. And in product marketing, there’s a lot of hype.
The truth is simpler: AI is a tool that lets me move faster on specific, bounded tasks. It’s not going to replace the parts of product marketing that matter most. But it’s reshaping how I work, and if you’re not paying attention to where, you’re probably missing something.
Let me walk through what’s actually working.
Synthesis at scale
The biggest shift for me has been in customer research synthesis. I used to spend hours listening to call recordings, reading support tickets, reviewing customer interviews. The task wasn’t intellectual. It was mechanical. Listen, take notes, find patterns, repeat.
Now I feed call transcripts and ticket threads into an AI tool and ask it to surface patterns. Not insights yet. Patterns. What problems show up again and again? What language do customers use? Where do they get stuck?
This saves me easily ten hours a week. More important, it surfaces things I would have missed by listening to a subset of calls. When you can process fifty call transcripts instead of five, the weak signals get louder.
The catch: I still have to synthesize the synthesis. The AI might tell me “customers mention integration complexity seven times,” but it won’t tell me why that matters for positioning or if it’s really a blocker or just friction. That’s the part where I earn my salary.
Competitive intelligence that doesn’t “lie”
Competitive intelligence is the use case where people oversell AI the hardest. “Monitor your entire competitive set continuously!” they say.
Reality is messier.
Here’s what I actually do: I have Claude scan our competitors’ websites, blog posts, and job postings weekly. It surfaces changes. New hires suggest product direction. Job descriptions telegraph features they’re building. A refresh of their pricing page matters.
But I don’t trust the interpretation. I read the source material myself. The AI is a filter. It tells me what to pay attention to, not what it means.
I’ve found it’s also useful for tracking how competitors talk about problems. I’ll collect their messaging artifacts, feed them to an AI, and ask it to extract claims and positioning patterns. Where do they lean into differentiation? How do they frame their buyer personas? What does their value prop hierarchy look like?
This doesn’t replace talking to customers and prospects, but it frames the conversation differently. When someone mentions a competitor, I actually know what the competitor has been saying, beyond what I remember or what sales passed along.
Content generation with human judgment
I write a lot. Product launches, sales enablement, thought leadership, internal alignment docs, even some of the creative brief language that goes to design (and this blog). I used to write drafts linearly. Start with an outline, write the opening, build from there.
Now I use AI to generate rough first drafts. Not because the AI is going to write something publication-ready. It won’t. But it gets me to forty percent done in five minutes, which beats staring at a blank page.
The process looks like this: I’ll write a detailed prompt that includes context, tone, the exact point I want to land on, constraints. I feed it to Claude (my preferred) or another LLM. It generates something. I read it, tear it apart, rewrite the weak parts, sometimes throw it out entirely and start from scratch.
What I don’t do is hand it to someone and say “clean this up.” That’s lazy. The AI-generated draft is a starting point, not a shortcut. The work is in the judgment. What’s actually true? What’s worth saying? What’s noise?
I’m seeing juniors make a mistake here. They treat AI-generated content as finished goods. It shows. The writing is competent but hollow. There’s no point. No perspective. No risk. It reads like the internal documentation of a competent but uninspired company.
I use AI for this the same way I use a research assistant. It does work that would be dull for me to do.
I keep the judgment part.
Message testing and positioning iteration
This is where I’m still learning. I’ve started using AI to generate positioning variations. Same core message, different framings. Then I’ll stress-test them.
Here’s an example: I’m positioning a feature as “time savings.” I’ll ask an AI to reframe it as “faster decision-making,” “competitive advantage,” “reduced operational risk.” Each reframing is the same thing, but it lands differently depending on who’s reading.
Then I test them. But here’s the gap in the current tools: the testing isn’t as good as talking to real prospects. I can run surveys through Typeform, and I do. But when I’m testing a really important positioning decision, I still get on calls. The AI helps me generate variations faster. The customer still tells me what actually lands.
What I’m curious about is the long tail. For routine content updates or lower-stakes messages, can I test and iterate entirely in AI? Early signs suggest yes, but it requires you to have good baseline assumptions about who your customer is and what they care about.
btw, I still use my message map framework.
The parts AI can’t touch
There’s a lot of product marketing work that AI doesn’t help with. Accepting that has been freeing.
Building cross-functional relationships is still a human job. Sales needs to trust that you understand their problems. Product needs to know you’re advocating for customers, not simply translating requirements. Design wants to know you’re thinking about the total user experience, not only the elevator pitch.
These relationships take time and repeated interaction. No AI tool can compress that.
Strategic judgment is another no-go zone. “What should our positioning actually be?” isn’t a question an AI can answer. It can help you gather data, surface patterns, generate options. But the decision about who you want to be in the market, where you’re actually defensible, what bets are worth making, that has to come from you.
I’m also skeptical of AI for customer empathy. You can feed an AI a thousand customer support tickets and it will tell you customers are frustrated. But empathy requires you to sit with that frustration. To feel it a little. To let it shape how you think about the product. That’s not a data processing problem. It’s a human one.
And I haven’t found AI useful for the really important internal communication. Launches need vision. Announcements need tone. Alignment conversations need someone willing to say hard things. These require you to take a position, and taking a position means being on the hook for it. An AI-generated memo never carries that weight.
Where I’m still figuring it out
I’m experimenting with using AI to help me think through strategy. Not generate strategy. Thinking through it. I’ll write out a positioning thesis, have the AI poke holes in it, force myself to respond. It’s like having a critical sparring partner who always has time and never gets tired of the conversation.
It works sometimes. Other times it’s busy work, a form of productive procrastination. I’m still trying to figure out when to trust it and when to just sit with my own thinking.
I’m also interested in if AI can help with internal storytelling. How do you take a complex product direction and make it emotionally compelling to the team? The data is never enough. You need narrative. Early experiments suggest AI is still too surface-level here, but the direction is interesting.
The real shift
The bigger picture isn’t that AI is replacing any particular part of product marketing. It’s that the parts that required raw processing power are getting cheaper and faster. That means the parts that require judgment, creativity, relationship-building, and taste are getting more valuable.
If you’re a PMM and you’re worried AI is coming for your job, you’re focused on the wrong thing. The mechanical parts were never why you got hired. You got hired to make good decisions about who the product is for and why they should care.
AI is automating the grunt work. That’s actually great for the profession, if we’re honest. It means less time in spreadsheets and more time thinking. Less time on first drafts and more time on strategy.
But it only works if you’re willing to do the judgment part. If you try to replace judgment with AI, you’ll get what AI gives you: competent, forgettable work.
The PMMs who’ll thrive in the next few years are the ones who use AI to move faster on the mechanical tasks and spend the time they save on the parts that actually matter. Not less work. Different work. Better work.
FAQ
What specific AI tools are you using?
I use Claude & Gemini Pro for writing and synthesis work. I have subscriptions to a few competitive intelligence tools that use AI under the hood. I experiment with other things occasionally. What matters more than the specific tool is understanding how to prompt well and knowing when to not use a tool at all.
Should I be using AI to create marketing collateral for sales?
Partially. AI is good at generating first drafts of sales decks, one-pagers, competitor battle cards. I’ve generated dozens of these. What I won’t do is hand a fully AI-generated collateral piece to the sales team without my fingerprints all over it. You’re putting your credibility on that material. Make sure it reflects your actual thinking.
If AI can do customer research synthesis, do I still need to do customer interviews?
Yes, but maybe not as many. If you have a lot of existing customer data (calls, tickets, feedback), AI synthesis can sometimes answer your tactical questions. Strategic questions still need real conversation. And you need to do interviews to maintain your intuition about the customer. That matters.
Thinking about how AI fits into your product marketing workflow? Let’s connect, always happy to talk through what’s working.
