Table Of Contents
The product manager: from author to editor
The product owner: the most exposed seat at the table
The designer: taste over screen-drawing
The product marketer: highest automation exposure, and a moving target
The pattern across the whole table
Our first piece looked at what generative AI is doing to software engineering, and found a consistent pattern: the work is moving up the stack, from producing code to directing, verifying, and owning it. That same force is now reshaping everyone else around the product table. Here’s the honest, fact-checked version of what’s happening to product managers, product owners, designers, and product marketers, and what it means for the people who do those jobs.
Before the role-by-role tour, one caveat worth planting up front, because it applies to all of them. Adoption is nearly universal but value capture is not. MIT’s Project NANDA reported that roughly 95% of enterprise generative-AI deployments produced no measurable profit-and-loss impact, and McKinsey’s research shows the same shape: almost every organization uses AI somewhere, but only a small minority has turned that usage into results. Every optimistic number below should be read against that backdrop. Using AI and benefiting from it are different achievements.
The product manager: from author to editor
Product management may be the role most obviously transformed, because so much of it was always synthesis and documentation, exactly what large language models do quickly. Reading hundreds of support tickets, survey responses, and interview transcripts to find the themes; drafting PRDs, user stories, and release notes; assembling competitive analyses; querying product analytics. AI compresses all of it. The first draft of nearly everything a PM produces can now appear in minutes rather than days.
What that compression does not do is make the decisions. This is the theme that recurs across every credible account: when generation gets cheap, the bottleneck moves from producing options to choosing among them. One product leader framed it well, arguing that PMs who mistake the speed of generation for the speed of decision-making will ship faster and ship worse. The judgment, deciding which market to serve, when to cut a feature, how to sequence a roadmap, still belongs to a human with organizational context.
The data reflects the shift in what’s valued. In Productboard’s survey of product leaders, roughly 59% named strategy and business acumen as the most important PM skills for the next few years, ranking them above execution mechanics. The center of gravity is moving from output to outcomes: less time producing artifacts, more time on discovery, customer contact, and prioritization.
Two structural changes are worth watching. First, the traditional ratio of one PM to a handful of engineers is loosening as AI absorbs coordination overhead. Second, roles are starting to blur into generalists: LinkedIn’s chief product officer described replacing the company’s associate-PM track with a “product builder” program that trains people across product, design, and engineering, a signal that the next generation may be expected to span disciplines rather than specialize early. If that pattern spreads, “PM” becomes a wider job, not a narrower one.
Table Of Contents
The product owner: the most exposed seat at the table
If the PM role is being elevated, the product owner role, in its classic Scrum definition, is the one facing the hardest questions. Strip the PO job to its core and you get backlog management: writing user stories, defining acceptance criteria, slicing large items into small ones, prioritizing the queue, and keeping requirements clear for the development team. That is a remarkably precise description of what generative AI is good at.
The tooling reflects this directly. Agile training bodies now teach POs to hand a feature description to an AI and get back a set of well-formed user stories, to auto-generate acceptance criteria and edge cases (the “expired session during an active API call” scenario a human might forget), and to cluster incoming requests into themes. Vendors advertise large reductions in story-drafting time. Whatever the exact figure, the direction is clear: the production half of the PO job is being automated faster than almost any other role on the team.
The honest implication is uncomfortable. In many organizations the PO role was already being questioned and, increasingly, merged upward into product management, with the PM owning the full lifecycle from discovery through release rather than handing an execution slice to a separate owner. AI accelerates that consolidation, because the coordination-and-documentation work that justified a dedicated seat is exactly the work getting absorbed. The POs who thrive will be the ones who move toward the parts machines can’t do: genuine prioritization under conflicting stakeholder pressure, the political capital to say no, and deep enough customer understanding to know which AI-drafted story actually matters. The ones who define themselves purely as backlog administrators are the most exposed people on the team.
Table Of Contents
The designer: taste over screen-drawing
Design has moved from cautious experimentation to daily dependence remarkably fast. The 2026 AI in Design report found that about 91% of designers now use AI for design tasks at least weekly, up from 54% a year earlier, and that the average designer’s toolstack more than doubled from three tools to seven, with the mix still churning as people hunt for their favorites. Tools like Figma Make and Figma’s AI agent now generate interface drafts from a prompt and push toward design-to-code handoff, collapsing steps that used to take hours.
But the design data is also the most usefully sobering, because designers have been unusually candid about the limits. In Figma’s own AI research, about 78% of designers and developers said AI boosts their efficiency, yet only 58% said it improves the quality of their work, roughly 40% said they don’t yet trust AI outputs enough to rely on them, and only about 27% believed AI would meaningfully move their company’s goals in the year ahead. That’s a workforce that finds the tools genuinely faster while remaining clear-eyed that faster is not the same as better.
The reason maps neatly onto the engineering story. Generating a plausible screen is a small fraction of design work; the hard part is continuity, remembering the previous trade-off, honoring the design system, covering the messy real states (permissions conflicts, expired sessions, irreversible actions), and making something a team can actually ship. AI without deep product context produces convincing surfaces; the human supplies the context, the taste, and the judgment about when the system should bend. Reviewers also note that AI-generated front-end code often falls short on accessibility and semantic quality, which is precisely the kind of gap only a knowledgeable human catches. The frequently repeated forecast, that designers shift from “screen decorators” to product strategists, is less a slogan than a description of where the remaining value concentrates.
Table Of Contents
The product marketer: highest automation exposure, and a moving target
Of the four roles, product and content marketing may face the most direct automation pressure, simply because so much of it is content production. The numbers are striking: surveys report that around 94% of marketers plan to use AI in their content process, and the share of marketers not using AI for blog creation reportedly collapsed from roughly 65% to about 5% in two years. The Marketing AI Institute found teams piloting or scaling AI rising to about 60% of respondents, up from 42% in 2023. On the advertising side, the IAB reported that 86% of media buyers use or plan to use AI to build AI-generated video ads, and separate research put AI image creation for display and social ads around 73% of US advertisers.
Anthropic’s research on AI’s labor-market impact is blunt about the exposure here: it identifies market-research analysts and marketing specialists among the occupations most exposed to AI, estimating that roughly 65% of current marketing tasks could be automated or meaningfully supported by AI. That is a high number, and it’s why “product marketer” belongs in any honest conversation about which roles change most.
Yet the more interesting shift for product marketers isn’t automation of the old job; it’s that the job’s terrain is being rewired underneath them. AI-generated answers are becoming a primary discovery layer. Google’s AI Overviews now appear on roughly half of searches, and a growing share of buyers research products through assistants like ChatGPT before they ever touch a brand’s site. That inverts a core marketing discipline: the goal shifts from ranking in a list of links to being understood and cited by AI systems, a practice the field has started calling answer-engine optimization. Content increasingly has to be structured so a model can interpret and reuse it, not just so a human can skim it.
Two constraints keep this from being a pure efficiency story. Trust is one: surveys find a majority of internet users worry that AI is making online content less trustworthy, roughly 56% in recent eMarketer-cited data, and a meaningful share of marketers now label work “created with AI” partly in response. Governance is the other: reporting suggests most organizations still lack formal AI safeguards in their vendor contracts and marketing workflows, which means the flood of cheap content is arriving faster than the oversight to keep it accurate and on-brand. For product marketers, the durable value is moving toward positioning, narrative, brand trust, and the strategic judgment about what’s worth saying, not the mechanical production of saying it.
Table Of Contents
The pattern across the whole table
Line up the four roles and the same shape appears every time, and it’s the same shape we found with engineers. The production half of each job, drafting, synthesizing, generating first versions and assets, compresses toward minutes. The judgment half, deciding what’s worth building, whether it’s any good, and who’s accountable for it, becomes the scarce and valuable part. The bottleneck moves from making things to verifying them. And roles blur, as the coordination work that separated them gets absorbed and generalist “builders” who can span disciplines become more attractive.
The macro forecasts fit this reframing rather than a simple replacement story. The World Economic Forum’s Future of Jobs research projects roughly 170 million new roles and 92 million displaced by the end of the decade, a large net gain paired with enormous churn, alongside a clear wage premium for people with AI skills. That’s not “the product team disappears.” It’s “the product team’s work is redistributed, and the mix of skills that pays off changes.”
Table Of Contents
What stays human, everywhere
Across all four roles, the same core survives contact with the tools. Deciding what is worth building. Understanding what people actually need versus what they said they wanted. Exercising taste and strategic judgment when the AI offers ten plausible options and only one is right. Holding stakeholder trust and organizational context that no model has access to. And being accountable when something ships and goes wrong. These are not content-generation problems, and there’s no evidence the current tools are close to owning them.
There’s also a shared hazard. Because AI makes everyone feel faster, and, as the engineering research showed, that feeling is an unreliable narrator, every role faces the temptation to confuse output with progress. A PM generating more specs, a marketer publishing more posts, a designer producing more screens, and a PO drafting more stories can all be busier and worse at the same time. The teams that win will measure outcomes, not activity.
Table Of Contents
How to prepare
The practical advice rhymes across the roles. Get genuinely fluent with the tools, then judge them by measured results rather than the sensation of speed. Invest disproportionately in the durable skills, strategy, taste, prioritization, customer understanding, and accountability, that appreciate as generation gets cheap. Treat every AI output as a draft to be reviewed by someone who owns the outcome, whether that output is a spec, a story, a screen, or a landing page. And expect role boundaries to keep softening; the people who can operate across product, design, and go-to-market will have room to move that specialists may not.
Table Of Contents
The bottom line
The story of generative AI on the product team isn’t replacement, and it isn’t hype-free transformation either. It’s the same quiet, consequential shift happening in every seat at once: the machine takes over more of the producing, and the human takes on more of the deciding, verifying, and owning. Engineers felt it first. Product managers, product owners, designers, and marketers are feeling it now. The teams that come out ahead won’t be the ones with the most tools. They’ll be the ones who kept their judgment firmly in the loop, and got faster only where speed actually made the product better.
Table Of Contents
Sources
Figures in this article draw on: MIT’s Project NANDA (share of enterprise GenAI deployments with no measurable P&L impact) and McKinsey’s State of AI (adoption vs. value capture); Productboard’s product-leader survey (top PM skills); public comments from LinkedIn’s chief product officer on generalist “product builder” hiring; agile training and industry write-ups on AI-assisted backlog and user-story workflows (illustrative capability claims, not independently audited); Figma’s 2025 AI report (efficiency, quality, trust and impact figures) and the 2026 AI in Design report (weekly-usage and toolstack figures; note: Anthropic, maker of Claude, is a disclosed partner of that survey, which was conducted independently); the Marketing AI Institute, IAB, Advertiser Perceptions and eMarketer (marketing and advertising adoption, AI Overviews reach, and consumer-trust figures); Anthropic’s research on AI’s labor-market impact (marketing-task exposure); and the World Economic Forum’s Future of Jobs research (job creation, displacement, and the AI-skills wage premium). Where a statistic originates from a vendor or a single survey, it is attributed as such; readers should consult the original reports for full methodology. Widely circulated numbers that could not be traced to a credible primary source were left out.