I've been a product leader for over two decades. I've shipped platforms at IBM, built AI portfolios at Teradata, held 22 patents, and taught product management as an adjunct professor for 18 years at NC State. And I can tell you that the most common question I hear from PMs today — at conferences, in my classes, in advisory sessions — is the same question, over and over.
"Is AI going to replace product managers?"
The short answer is no. The longer answer is that AI is going to replace a certain kind of product management — the kind that was mostly about coordination, status updates, and Jira tickets. If that's your job, you're right to worry. But if your job is about judgment, customer insight, and the hard work of making technology actually useful to human beings — AI doesn't replace you. It gives you a superpower you've never had before.
The job hasn't changed. The leverage has.
Product management has always been about the same thing: understanding a problem deeply enough to build something that solves it. That requires talking to customers. It requires synthesis — taking a mess of signals from sales, engineering, support, and the market and turning them into a clear direction. It requires the ability to say no to good ideas so you can ship the right one. None of that goes away when you add AI to the toolchain.
What changes is the speed at which you can move from insight to execution. AI compresses the cycle. You can prototype faster. You can analyze user data at a scale that used to require a full analytics team. You can generate product specs, test copy, run simulations, and evaluate trade-offs in hours instead of weeks. That's leverage.
But leverage without judgment is just velocity in the wrong direction.
The new skill stack
Here's what I tell my students and the teams I advise: the core competencies of product management haven't changed, but the skill stack has expanded. If I were writing the job description for a PM in 2026, it would include everything it did before — and three new capabilities on top.
AI literacy, not AI engineering.
You don't need to train models. But you need to understand what models can and can't do. You need to know the difference between a large language model and a fine-tuned classifier. You need to understand tokens, latency, hallucination rates, and why "accuracy" is a more nuanced metric than your stakeholders think it is. When your engineering team says "we can't guarantee the output," you need to understand why — and how to build a product around that uncertainty instead of pretending it doesn't exist.
Prompt design as product design.
For an increasing number of products, the spec is the prompt. The way you instruct the model — what context you give it, what constraints you set, how you chain calls together — directly determines the user experience. PMs who understand this are building better products than PMs who treat the model as a black box and hand the prompt to engineering. System prompts are product decisions. They should be owned accordingly.
Evaluation as a first-class discipline.
Traditional products are deterministic: you build a feature, it works or it doesn't. AI products are probabilistic: they work most of the time, and the question is whether "most" is good enough. This means evaluation — how you measure whether your AI is performing well — becomes as important as the feature itself. PMs need to define what "good" looks like for their specific use case, build evaluation criteria that reflect real user outcomes, and create feedback loops that catch regression before customers do.
The uncomfortable truth
The PMs who will thrive in the age of AI are not the ones who learn to use AI tools the fastest. They're the ones who already had the hardest-to-automate skills: customer empathy, cross-functional judgment, the ability to make irreversible decisions under uncertainty, and the discipline to kill projects that aren't working. AI amplifies those skills. It doesn't create them.
What I've seen go wrong
In my advisory work, I see the same failure mode over and over: teams that treat AI features like traditional features. They write a PRD, throw it over the wall to engineering, launch it, and wonder why customers are confused, or worse, distrustful. The problem is that AI products require a different kind of product ownership — one where you stay close to the output, watch how users interact with uncertainty, and iterate on the model behavior the way you would iterate on a UI.
The other failure I see is PMs who outsource their thinking to AI. They use ChatGPT to write their strategy docs, their competitive analyses, their OKRs. And the output is perfectly fine — and perfectly generic. The value of a PM is not in the document. It's in the judgment that went into it. If you're using AI to generate your thinking instead of accelerating it, you've made yourself replaceable.
The opportunity is bigger than most PMs realize
The shift to AI-native products is the largest platform transition since mobile. And just like mobile, the companies that win won't be the ones with the best technology — they'll be the ones with the best product thinking. Someone has to figure out how to make these capabilities useful, trustworthy, and valuable to real humans in real workflows. That's the PM's job. It always has been.
What excites me most is this: we're in a moment where the product manager's core skill — connecting technology to human need — has never been more important. The technology is moving faster than it ever has. The models are more capable than most organizations know what to do with. The gap between what AI can do and what it should do is where product managers live. And that gap is getting wider, not narrower.
Strategy without building is just a slide deck. I'd rather ship.
The age of AI doesn't make product management obsolete. It makes it essential. But only if you're willing to evolve — to learn the new stack, stay close to the technology, and keep doing the hard work of understanding what people actually need.
That's the job. It always was.