With vibe coding, you can build an app in a single day. AI writes the code for us, helps us write, and even drafts the analysis reports. Our capacity to make things has exploded.
There's no point in putting up a hundred buildings that sit half empty.
Publishing is no different. Once AI started helping with the writing, books could be produced at speed—until we arrived at the almost laughable spectacle of "vibe publishing." In the rush to churn things out, we skip the one question that matters: who is any of it actually for?
That question lands squarely on the essence of the product manager's job.
What a PM Actually Does
A product manager is responsible for a product's entire arc—strategy, planning, development, launch, and growth. People often call the PM "the CEO of the product," and the phrase is surprisingly accurate.
In concrete terms, the work looks like this. You set the product's long-term direction and lay out the staged goals—the roadmap. You define features, set priorities, and write user stories so the development team gets a clear specPRD, Product Requirements Document. You analyze user feedback, market trends, and product data to find where to improve. And you reconcile opinions among developers, designers, and marketers, driving the decisions.
Every PM carries three axes in their head at all times. Can people actually use itUsability? Can it be builtFeasibility? Does it hold up as a businessViability? Weighing all three at once is a PM's daily reality.
The PM's Three Axes of Judgment
It's easy to confuse the role with adjacent ones, so here's the short version. The PM focuses on the product's strategy and vision. The PO (product owner) owns backlog priorities and outcomes within an agile team. The service planner concentrates on feature implementation and user-flow design. And the project manager manages schedule, cost, and risk.
But AI Now Handles a Big Share of the PM's Work
This is where the trouble begins.
Data analysis? AI does it faster and more accurately. Market research? A competitor analysis lands in minutes. Drafting a PRD? One prompt and a credible document appears. User stories, A/B test designs—AI can help with all of it.
Seen through the framework in Min Yeon-gi's book Augmented Human, this is the same story as the shift from Work 1.0 to Work 3.0. We moved from an age when skilled labor itself was competence (Work 1.0) to the age of the knowledge worker who extracts information from data (Work 2.0), and we have now entered an age in which AI stands in for both skill and knowledge (Work 3.0). Of competence's three axes—Skill, Knowledge, and Attitude—two are now being replaced.
The Three Axes of Competence
PMs are no exception. The abilities long held up as the core of the job—data analysis, technical grasp, UX design—are now largely within AI's reach.
So What's Left for the PM?
Augmented Human tells the story of the magic mirror from Snow White. The stepmother asks, "Who is the fairest of them all?" and the mirror answers. But the mirror has never once wrestled with what "fair" even means. It simply hands back an answer anchored in the biases of past data. Even for a question with no right answer, the mirror gave one—right or wrong—and when the stepmother didn't hear what she wanted, she smashed it.
The PM's situation looks much the same. Ask AI, "Should we add this feature?" and a plausible, data-backed answer comes out. Ask, "What strategy are competitors running?" and you get a tidy analysis. But AI cannot pose to itself the question, "Why should this product exist in the first place?"
It's the same truth I confirm every day in publishing. However fast AI generates content, choosing the subject, defining the reader, and designing a book's direction is work that has to be done by a person. A book isn't a slab of printed paper that files away an author's knowledge—it enters a reader's life and fills the intellect. You need someone asking what to put inside that building, and what change it will set in motion.
Products are the same. In the end, three things remain the PM's.
First, the disposition to judge value. Deciding whether what AI produced truly means something to users, and what reason this product has to exist in the market. Weighing the three axes—usability, feasibility, viability—is still a person's job. Especially when those three collide—when users want it, but it's technically hard and commercially uncertain—deciding which way to lean is something AI cannot do.
Second, the disposition to keep learning. Markets change constantly. There's no guarantee that yesterday's data still holds for today's decision. AI is locked inside past data, but a person can sense the early signs of change that haven't yet become data. That is exactly why a PM has to see the field firsthand, meet users, and read the air of the market.
Third, the disposition to connect and expand. A PM is the person who reconciles views among developers, designers, and marketers. This is not a simple communication skill. It is the work of connecting different perspectives and pulling them into a single direction—the product. Someone who connects across existing boundaries and holds a direction for doing so—this is exactly what Min Yeon-gi means by the augmented human.
A PM Is Not Simply Someone Who Uses AI Well
Here's where it's easy to get the wrong idea.
"So a PM who makes good use of AI is the one who survives, right?"
No. As Min Yeon-gi puts it precisely, the augmented human is not someone who uses AI well. It is someone who judges whether AI's output is enough, who never stops learning, and who holds a direction that connects beyond existing boundaries.
For a PM, it comes down to this. Not the person who takes AI's draft PRD as-is, but the person who spots the context missing from it. The person who reads, in the data AI analyzed, the story the numbers don't tell. The person who, faced with the feature list AI proposed, can ask, "Why are we building this?"
We live in an era when forging a connection with the reader is harder—and more important—than mindlessly writing and piling up more pages. Products are no different. Designing a product that genuinely enters a user's life is harder, and matters more, than mindlessly stacking up more features.
What Will You Ask the AI?
In the Work 3.0 era, the PM role won't disappear. If anything, it grows more important. What changes, fundamentally, is the substance of the work—from a PM who analyzes data and writes documents to a PM who judges direction and designs value.
The question Augmented Human leaves us with comes down to this.
What will you ask the AI?
And in front of the answer it gives back, what will you judge?
The PM who holds their own answer to that question is this era's augmented human.
