Back in 2022, I wrote The Role of a Product Manager, and two years later I wrote an article for Scotia Digital’s Future Fest Craft Magazine with six predictions about how the role would change over the next decade. In just two years, many of those predictions have already come true. I was too conservative framing this as a ten-year horizon. The pace is accelerating so fast that predicting the next ten years feels naive. What we can speak to with confidence is now and the next twelve to twenty-four months.
Looking back at that piece, I predicted AI-powered assistance, the rise of the product visionary, deeper customer-centricity, and the importance of adaptability. Reasonable stuff. Forward-looking enough.
I got several of those right. But I missed some critical shifts, or at least failed to articulate them clearly enough. One of the most important: product management has become the bottleneck.
“The bottleneck is no longer engineering. It’s product management.”1
Andrew Ng claims his teams can now build in a weekend what used to take six engineers three months. The constraint has shifted from can we build this? to what exactly should we build, and should we build it at all? Some AI-native teams are moving toward a ratio of one product manager per engineer — the exact inverse of the traditional 6:1 model.
When I wrote my 2022 post on the role of a PM, the premise was straightforward: the PM is the bridge between business, design, and engineering. They own the “what” and the “why.” That was true then. It’s still true now.
But the stakes attached to that “what” have never been higher.
What the Role Was
The traditional PM definition hasn’t really changed. You own the product vision. You synthesize customer needs, business goals, and technical constraints into something buildable. You hold the four product risks: value, usability, feasibility, and business viability, simultaneously in your head and make tradeoffs accordingly.
Marty Cagan described this well in Inspired: the PM is responsible for a product that customers love and that works for the business. Not one or the other.
What made this hard, and what justified a dedicated PM role, was the coordination overhead. Engineering was expensive and specialized. Design and engineering rarely overlapped. Market research took weeks. Decisions required buy-in from multiple stakeholders, each with different information. The PM bridged those gaps.
That model worked. But it had a shadow side that Cagan recently called out directly: “product management theater”2: roles that go through the motions of product work (writing specs, facilitating planning sessions, creating decks) without genuinely owning product outcomes. A lot of PM jobs were, if we’re honest, coordination jobs in disguise, not to mention feature factory PMs who operate more like delivery managers with one goal: ship a lot of features, on time.
When I wrote my 2024 article, I framed the coordination role as something that would evolve. I underestimated how fast “evolve” would mean “compress toward irrelevance.”
What My 2024 Predictions Got Right — and What I Missed
Looking back at my six predictions from 2024, I can be fairly honest about the scorecard.
What I got right:
The prediction about AI-powered assistance has come true faster than I imagined. AI now synthesizes research, drafts PRDs, generates user personas, summarizes customer interviews, and writes competitive analyses in minutes. What I framed as a future collaboration is already the present baseline.
The need for AI fluency that I called “AI-savvy PMs” has also accelerated beyond what I described. I talked about a “baseline understanding of ML concepts.” That was too modest. Today’s AI PM needs to understand model failure modes, design evaluations, think through responsible AI risks, and work directly with agentic systems. The bar is higher.
The intensifying focus on customer-centricity holds. In a world where anyone can spin up a product quickly, deep user empathy remains a genuine moat.
What I missed:
I predicted “the rise of the product visionary”: PMs who anticipate future trends and translate them into compelling roadmaps. That’s not wrong exactly, but it’s too abstract for what’s actually needed. Visionaries who operate on 6–12 month cycles are too slow. What’s emerged instead is something more operational: product taste. The ability to know what’s good before users tell you, evaluate it quickly, and move.
I also underestimated how much the boundaries between roles would dissolve. I mentioned cross-functional collaboration as a growing skill. What’s actually happening is more structural: design, PM, and engineering are converging into the same person at AI-native teams. LinkedIn recently replaced its Associate PM program with a “Product Builder” program that trains people across product, design, and engineering simultaneously. That’s not an incremental shift. That’s a reimagination of the role.
And I missed the Andrew Ng framing entirely. I didn’t see that the PM would become the bottleneck.
What AI Is Disrupting Right Now
The Information Mover Is Obsolete
Nikhyl Singhal,3 a career advisor and former Facebook product leader, offers a framework I keep coming back to: information movers vs. builders.
The information mover is the PM who moves PowerPoints from one team to another, synthesizes what the customer said in a research session, writes up the PRD, and aligns stakeholders. That person, he argues, is becoming obsolete.
“The information mover is going to be a dinosaur,” Singhal says. “The builder — the person who actually thinks and creates — there’s a renaissance coming for them.”
AI can now do the information movement. Not as well as a skilled PM, perhaps. But well enough, fast enough, to make that skill insufficient as a value proposition.
The uncomfortable implication: many PMs who have been “successful” are successful for the wrong reasons. If your edge was synthesis, documentation, or coordination: that edge is eroding now.
The Iteration Cycle Has Collapsed
Cat Wu,4 Head of Product for Claude Code at Anthropic, captures this in one sentence: “What used to take 6 months now can take 1 day.”
Feature development cycles that once required quarters of planning, sprint estimation, and stakeholder alignment can now ship in days. This doesn’t just change how teams operate; it breaks the premise of annual roadmapping, multi-quarter planning, and the PM’s traditional role as the long-horizon thinker.
Jenny Wen,5 Head of Design for Claude at Anthropic, describes the same compression from the design side: “Vision used to be 6 months out. Now it’s 3 to 6 weeks. If you can’t let go of your old vision and build a new one fast, you’ll always be behind.”
The Roles Are Converging
When engineers can ship features without traditional specs, and designers can write code directly, the traditional handoff model (PM writes PRD, designer creates mockups, engineers build) starts to break down.
Jeetu Patel,6 CPO at Cisco, puts it bluntly: “Fast-forward yourself 6 months. Whatever you’re doing today, assume AI can do it. Now ask: what’s left?”
What’s left, in most cases, is judgment. Taste. The decision about which of the ten AI-generated options is the right one. The story that convinces the organization to move.
What the Role Is Becoming
The PM as Judgment Engine
The scarce resource in product development is no longer execution speed; AI provides that. It’s the quality and speed of decisions.
Andrew Ng’s teams are “increasingly relying on gut” for decisions that previously required weeks of A/B testing. He explicitly calls A/B testing “one of the slowest strategies in our portfolio.” When you can prototype in a day, waiting a week for statistical significance becomes a competitive disadvantage.
This is what the PM role is converging on: someone who can make high-quality decisions quickly, under uncertainty, with incomplete data, and be right often enough to matter.
Keith Rabois,7 the venture capitalist and former PayPal and Square executive, is blunter than most: “The idea of a PM as traditionally defined makes no sense in the future. What survives is CEO-like judgment — what are we building and why?”
The PM as Product Taste-Maker
Cat Wu4 frames her #1 requirement for the PM role as product taste: “the ability to know what’s good before users tell you.”
I find this framing more precise than “vision.” Vision suggests you’re pointing at a 12-month horizon. Taste is operational: it’s what you do when AI hands you ten options and you need to pick one in five minutes.
Aakash Gupta describes a useful flywheel: the PM who evaluates 15 prototypes a week builds taste faster than the PM who reviews one spec a month. Speed creates compounding judgment. The “taste premium,” as he puts it, doesn’t depreciate in the AI era; it compounds.8
The PM as Responsible AI Owner
This is the part of the role that I think is least discussed and most consequential.
Marty Cagan wrote recently that “most product managers will be expected to understand how the enabling AI technology works, what the range of risks involved are, and the work required to mitigate them.”9 McKinsey data shows demand for AI fluency in job postings has grown nearly sevenfold in two years, and much of that demand explicitly includes responsible AI competencies.
When your product includes AI-powered recommendations, automated decision-making, or generative outputs, the PM owns the failure modes. Not the legal team. Not the ethics committee. The PM who shipped it.
This includes understanding where models hallucinate, how bias can enter training data, what transparency means for your users, and when to pump the brakes on a feature that technically works but shouldn’t be deployed.
This is new. It wasn’t in any PM job description in 2022.
What Doesn’t Change
Here’s what I’m confident about: the fundamentals of the PM role are not going away.
The “what and why” still lives with the PM. AI generates “how” options with unprecedented speed and volume. The PM still decides which “what” to pursue.
User empathy remains irreplaceable. AI can analyze thousands of support tickets. It cannot replicate, at least not yet, the insight from watching a frustrated user try to complete a task for the first time.
Storytelling (Jeetu Patel again): “The ability to tell a story that moves an organization is not automatable.” As AI compresses execution, the organizational bottleneck is increasingly alignment. The PM who can build conviction across engineering, design, data, legal, and executive stakeholders becomes more valuable, not less.
And the four product risks: value, usability, feasibility, business viability. None of these have changed. AI just changes the tools used to address each one.
Specializations Are Being Rewritten
The PM specializations I described in 2022 (Technical PM, Data/AI PM, Growth PM, Enterprise PM, Startup PM) are all still real, but they’ve been substantially rewritten.
Technical PM → Model PM. Understanding APIs and system design is still relevant. But the technical PM of 2026 needs to understand AI systems specifically: how evals work, what agentic architectures look like, where models fail and why.
Data/AI PM → AI Product PM. Now one of the most in-demand PM roles in tech. The distinct skills are evaluation design, model capability awareness, and responsible AI oversight.
Growth PM → Activation PM. The hardest problem in AI products isn’t acquisition; it’s getting users to their first moment of genuine value fast enough. AI-automated growth workflows are changing tactics, but judgment about what to automate and for whom remains human.
Enterprise PM → AI Transformation PM. Enterprise PMs are now leading projects that replace entire workflow layers, not just adding features. Navigating organizational change, compliance, and trust-building with skeptical enterprise buyers is the new enterprise PM superpower.
Startup/Founding PM → Builder PM. The lean PM who can span design, build, and ship, now amplified by AI, is the most powerful individual contributor in tech. Andrew Ng’s prediction of the “one-person billion-dollar company” is being enabled by exactly this archetype.
The Stakes Are Getting Higher
I want to be direct about what this means for the profession as a whole.
If you’re in a PM role where most of your time goes to documentation, status updates, facilitating planning sessions, and synthesizing information that other people will act on: that’s a role under pressure. Not in five years. Now.
If you’re early in your career, the traditional junior PM path, where you learn the craft by doing the coordination work, is narrowing. The entry-level tasks that used to serve as apprenticeship are automating. LinkedIn’s replacement of its APM program with a “Product Builder” program is the institutional sign of this. What gets you a seat at the table now is proof of taste and judgment, not proof of process compliance.
Ant Murphy’s 2026 survey of product managers found that 59% rank strategy and business acumen as the most important skills for the next two to three years.10 Not communication. Not execution. Not collaboration. Strategy. The profession knows what’s coming.
But here’s what I believe, and what I’ve seen: for PMs who lead with judgment, taste, and genuine curiosity about what users need, this is the most exciting time to be in this role. The coordination overhead is lifting. What remains is the part that was always the point.
“The builders — the people who actually think, who create, who have opinions,” Singhal says, “there’s a renaissance coming for them.”3
I believe him. I’m building for it.
What shift in the PM role are you seeing most clearly in your own work? I’d love to hear how your team is adapting.
This post is the first in a two-part series updating my 2022 writing on product management. The second post covers the specific skills that matter most for PMs in the AI era.
-
Andrew Ng on the PM bottleneck: Hackernoon ↩
-
Marty Cagan, Product Management Theater,” Silicon Valley Product Group via Lenny’s Newsletter ↩
-
Nikhyl Singhal, “Why half of product managers are in trouble,” Lenny’s Podcast ↩ ↩2
-
Cat Wu, How Anthropic’s product team moves faster than anyone else,” Lenny’s Podcast ↩ ↩2
-
Jenny Wen, The design process is dead. Here’s what’s replacing it,” Lenny’s Podcast ↩
-
Jeetu Patel, AI is critical for humanity’s survival,” Lenny’s Podcast ↩
-
Keith Rabois, Hard truths about building in the AI era,” Lenny’s Podcast ↩
-
Aakash Gupta, “There’s a New PM Skill. It’s Called Taste at Speed” ↩
-
Ant Murphy, How Product is Changing in 2026” ↩
Comments