I spent three days last quarter writing a positioning document for a product launch. Interviewed customers, analyzed competitors, crafted messaging hierarchies, debated word choices with the team. The final output was solid—good enough to drive a $2M launch without major revisions.
Last week I fed ChatGPT the same inputs and asked it to write the positioning doc. It took 30 seconds and produced something 80% as good as my three-day effort.
That moment crystallized something I'd been avoiding: AI isn't coming for product marketing. It's already here. The question isn't whether it will change our jobs, but which parts survive and which parts disappear.
I spent the next month testing AI tools across every PMM workflow I run. Competitive intelligence gathering, messaging frameworks, launch briefs, sales enablement content, customer research synthesis. I wanted to know what AI could actually do versus what the hype promised.
The results surprised me. AI didn't replace the strategic thinking I expected it would struggle with. Instead, it eliminated the execution work I thought was safe from automation.
What Gets Automated Faster Than You Think
The positioning doc experience sent me down a rabbit hole. If AI could draft positioning in 30 seconds, what else could it handle?
I started feeding it competitive intelligence gathering tasks. Point an AI agent at three competitor websites and ask it to extract positioning, pricing, target customers, and key differentiators. It took four minutes and produced a comparison table that would have taken me two hours of manual research.
Then I tried messaging framework development. I gave it our value props, customer pain points, and competitive context. It generated a complete messaging hierarchy with primary message, supporting pillars, proof points, and persona-specific variations. The first draft needed refinement, but it was 70% there in two minutes versus the full day I'd normally spend.
Sales battlecards were next. I fed it our product specs, competitive intel, and recent win-loss interview notes. It produced objection handling scripts, competitive positioning traps, and discovery questions. Again, not perfect, but good enough to be dangerous.
The pattern became clear: AI excels at synthesis and formatting. Give it unstructured inputs and a clear output format, and it will produce decent first drafts faster than any human can. It doesn't get tired. It doesn't procrastinate. It doesn't need three rounds of stakeholder feedback to start writing.
What it can't do—yet—is make the judgment calls that separate good positioning from great positioning. It can't tell you which competitive differentiator will resonate most with your ICP. It can't feel the difference between messaging that sounds clever and messaging that converts. It can't read a room during a sales call and adjust the pitch on the fly.
But here's what keeps me up at night: the gap between what AI can do today and what it will do in 18 months is shrinking faster than most PMMs realize.
The Work That Stays Human
After a month of testing AI across my entire workflow, three types of work emerged as stubbornly human:
First, customer insight extraction. AI can transcribe interviews, summarize themes, and identify patterns in customer feedback. What it can't do is hear the pause before a customer answers a question and know that's where the real insight lives. It can't detect when someone is being polite versus honest. It can't push back on surface-level answers to uncover the uncomfortable truths that drive product decisions.
I ran an experiment. I had AI analyze ten customer interview transcripts and generate insights. Then I reviewed the same transcripts myself. AI caught the explicit themes people stated directly. It missed every single insight that required reading between the lines—the moments when customers contradicted themselves, the features they claimed to want but never used, the problems they minimized because they'd learned to work around them.
Customer insight work that drives actual product and GTM decisions still requires human judgment. AI can help organize and surface the data, but it can't replace the pattern recognition that comes from doing 200 customer interviews and learning to hear what people don't say.
Second, strategic positioning choices. AI can generate positioning options. It can't tell you which one will work in your market.
I tested this by giving AI the same inputs I'd used for real positioning projects, then comparing its recommendations to the choices we actually made. AI consistently suggested the safe, obvious positioning. It pattern-matched to what successful companies in our category had done before.
The positioning choices that actually worked—the ones that helped us win competitive deals and expand into new markets—required understanding context AI didn't have. Market timing, competitive vulnerabilities, internal capabilities, sales team strengths, customer buying patterns. AI could analyze each factor individually but couldn't weigh them against each other to make the strategic bet.
Third, stakeholder management and influence. AI can write the perfect exec brief. It can't navigate the political dynamics of getting three executives to agree on strategy when they each have different incentives.
Product marketing isn't just about creating artifacts—it's about building consensus across functions that naturally conflict. Product wants different things than sales. Marketing wants different things than customer success. Getting everyone aligned on positioning, pricing, or launch strategy requires understanding who has power, what they care about, and how to frame recommendations in ways that address their concerns.
I can't outsource that to AI any more than I can outsource building relationships with customers.
How PMM Work Actually Changes
The real shift isn't AI replacing PMMs—it's AI changing what PMMs spend time on.
Before AI: I spent 60% of my time on execution (writing docs, formatting decks, researching competitors, synthesizing customer feedback) and 40% on strategy (deciding what to position, choosing which markets to target, aligning stakeholders on direction).
With AI: that ratio flips. AI handles 80% of the execution grunt work, leaving me to spend 80% of my time on judgment calls, strategic decisions, and stakeholder alignment.
This sounds like a productivity win, and in some ways it is. I launched three products last quarter versus the one I could have handled manually. But it also changed the skill profile required to succeed.
The PMMs who thrive in an AI-augmented world aren't the ones who write the best first drafts—AI will out-write them every time. They're the ones who can look at ten AI-generated positioning options and immediately spot which one will resonate with their ICP. They're the ones who can take rough AI-generated competitive intelligence and know which insights matter and which are noise. They're the ones who can use AI to generate 80% of a sales deck in five minutes, then spend their time on the 20% that actually influences deals.
I started testing consolidated platforms like Segment8 that layer AI capabilities on top of core PMM workflows. The promise was compelling—instead of gluing together six different AI tools for competitive intel, messaging, and launch management, use one platform that understands the full PMM workflow. Update competitive positioning once, have AI propagate it across battlecards, sales decks, and messaging docs automatically. The time savings weren't just from AI—they were from eliminating the manual reformatting and copy-paste work between tools.
The shift I'm seeing is PMMs moving from makers to editors and strategists. AI makes the first draft. PMMs decide if it's the right draft, refine it based on context AI doesn't have, and navigate the stakeholder dynamics required to get it adopted.
The Skills That Matter Now
If AI handles execution and humans handle judgment, the skill gap becomes obvious: most PMMs spent their careers getting good at execution and adequate at judgment. The future requires flipping that ratio.
I've been deliberately practicing the skills that AI can't replicate:
Understanding customer psychology at a deeper level than what they say in interviews. Reading body language in video calls. Hearing what they don't say. Detecting patterns across 100 conversations that reveal truths no single interview would surface. This is the insight work that drives positioning AI can't match.
Making strategic bets with incomplete information. AI is trained on what worked in the past. Breakthrough positioning requires betting on what will work in the future despite limited evidence. That requires conviction AI doesn't have.
Building influence across functions. Getting product to prioritize features that support your positioning. Getting sales to adopt your messaging instead of reverting to what's comfortable. Getting executives to invest in the market segments you've identified. AI can write the business case, but it can't read the room and adjust in real-time.
Prompt engineering and AI tool evaluation. As counterintuitive as it sounds, succeeding with AI requires understanding how it works well enough to know when to use it and when not to. Knowing which tasks to automate and which to keep human. Knowing how to structure prompts that generate useful outputs versus garbage.
I'm not worried about AI eliminating product marketing as a function. I'm worried about it eliminating product marketers who can't adapt to working with AI as their primary productivity tool.
What This Means for Your Career
The uncomfortable truth: if your primary value is being good at creating positioning docs, competitive battlecards, or messaging frameworks, AI will compress your leverage over the next 18 months.
The PMMs who survive and thrive will be the ones who use AI to increase their output 5x while focusing on the judgment calls that drive revenue. They'll launch five products instead of one. They'll manage competitive intelligence for ten competitors instead of three. They'll produce sales enablement content for twenty use cases instead of five.
But they'll do it by becoming editors and strategists rather than makers. They'll review AI outputs instead of creating first drafts. They'll make strategic positioning choices instead of wordsmithing messaging all day. They'll spend time with customers extracting insights AI can't surface instead of summarizing survey data.
This isn't the end of product marketing. It's the evolution from execution-heavy work to judgment-heavy work. The question isn't whether AI will change your job—it already has. The question is whether you're developing the skills that matter in an AI-augmented world or clinging to the execution work that's already being automated.
I'm betting on judgment over execution. On customer insight over content creation. On strategic thinking over tactical production. Because those are the skills AI will struggle to replicate for the next decade, while the execution work I used to do manually gets automated over the next 18 months.
The PMMs who make that shift now will define what product marketing looks like in 2030. The ones who don't will spend the next five years wondering why their role feels increasingly automated.