Product manager openings just hit a 3-year high. Over 7,300 roles globally, up 75% from the 2023 low (Lenny Rachitsky / TrueUp, 2026). At the same time, OpenAI runs with fewer than 30 PMs (Aakash Gupta, 2026). Both of these things are true.
The resolution: two different PM roles are on the market right now, and most interviews only test for one of them.
What Happened to the PM Role?
The old PM job was think and spec. Write a PRD, negotiate priorities, present a roadmap, facilitate between engineering and design. That role still exists at companies that haven't caught up. It is expanding, actually. More openings, more headcount.
But the companies setting the pace went somewhere else entirely. LinkedIn scrapped its Associate Product Manager program in late 2025 and replaced it with the Associate Product Builder program. No resume required. Applicants submit a 60-second demo of something they built and answer how they used AI (Tomer Cohen, LinkedIn CPO, Dec 2025). The name change is the tell. Product manager to product builder. Not semantics. A job description rewrite.
Tomer Cohen said on Lenny's Podcast: "70% of the skills needed for jobs will change by 2030." LinkedIn is not waiting for 2030.
Can Your PMs Actually Build?
Lenny Rachitsky's survey of 1,750 PMs found the top AI use cases: writing PRDs (21.5%), creating prototypes (19.8%), improving communication (18.5%) (Lenny's Newsletter, 2025). Prototyping is already the second most common thing PMs use AI for. That number was close to zero two years ago.
94% of enterprise PMs now rely on AI tools daily (Productboard, 2025, n=379). PMs report saving an average of 4 hours per task (Productboard AI in PM Report, 2025). Non-engineers want to ship working software, and the tooling now lets them.
The job changed. Most companies just haven't updated the interview for it.
A PM who can prototype in 45 minutes with AI tools and validate whether the output is production-ready is a different hire than a PM who writes a spec and waits for sprint planning. The first one compresses the feedback loop from weeks to hours. The second one is overhead that hasn't been reclassified yet.
Why Most PM Interviews Test the Wrong Skills
Every PM interview I see still asks "how would you prioritize this roadmap?" or "walk me through a product you launched." Behavioral questions. Hypotheticals.
Nobody asks: build me a prototype in 45 minutes. Here are AI tools. Ship something.
Stripe already weaves AI prototyping into its PM interviews, and Google's AI teams test product sense with hands-on building exercises (Aakash Gupta, AI Product Sense Guide, 2025). Candidates get access to AI tools and a real problem. The format tests whether they can build, not whether they can present a slide deck about building.
Ant Murphy put it well: "Speed without direction is just chaos in disguise." The interview needs to test both. Can this person build fast? Can they also tell you why what they built matters, and what they would cut?
Meanwhile, 95% of enterprise generative AI pilots fail to deliver measurable ROI (Ant Murphy / Industry Research, 2025). The failure is judgment. PMs who can operate AI tools but cannot evaluate whether the output solves a real problem are expensive liabilities. The interview has to catch this.
Where Did the Entry-Level PM Ramp Go?
Entry-level PM roles at startups collapsed. Junior openings dropped sharply even as mid-sized companies increased them by roughly 243% (Lenny Rachitsky / Productify, 2025). The apprenticeship model that trained the last generation of PMs is breaking.
The ramp that used to exist, junior role to associate to senior, is disappearing at exactly the companies where you would learn to build with AI. Nate at OpenAI described their approach: fewer PMs, each one a generalist who can build alongside engineers. That is the direction. Not more PMs, better PMs.
What This Means for Hiring PMs
If the PM job now requires prototyping and shipping working software, you cannot assess that with behavioral questions. You need to watch them build. See how they scope a problem with AI tools available. See whether they validate output or accept it blindly.
This is the same logic that applies to engineering hiring. Capture the process, not just the output. Fairground does this for engineering evaluation today, and the same principle applies to PM hiring. Async work samples, live collaborative sessions, scored dimensions that reflect how someone actually works, structured feedback loops, all captured in one place. Start free, 100 credits, no sales call at fairground.work.
Related: Hiring Harness Engineers in 2026
Get started with Fairground in just few mins.
Plug and Play. Works well with your existing ATS.
100 Free Credits
