← Back to Blog

Why Half of Product Managers Are in Trouble: The AI Reckoning Nobody Saw Coming

• product-management, ai-products, career-development, leadership, product-strategy, tech-trends, skill-development, future-of-work

Why Half of Product Managers Are in Trouble: The AI Reckoning Nobody Saw Coming

I've been building AI products for the past three years, and I'll be honest: watching the product management discipline unravel in real-time is both fascinating and terrifying.

Nikhyl Singhal, who's led product teams at Meta and Google, recently dropped a truth bomb that sent ripples through the PM community: half of product managers are in trouble. Not struggling. Not adapting slowly. In trouble.

Having worked alongside hundreds of PMs and built teams from scratch, I can tell you he's not exaggerating. The ground is shifting beneath our feet, and most product managers are still using a playbook written for a world that no longer exists.

Let me break down what's actually happening, why it matters, and—most importantly—what separates the PMs who will thrive from those who'll become obsolete.

The Comfortable Era Is Over

For the past decade, product management has been one of the most coveted roles in tech. Six-figure salaries, strategic influence, the promise of building products millions use. The path was clear: learn some frameworks (RICE, anyone?), run A/B tests, write specs, coordinate engineering and design, ship features, rinse and repeat.

This model worked beautifully in a relatively stable environment where:

Then generative AI arrived and detonated the entire foundation.

Singhal's observation isn't about AI replacing PMs—it's about AI fundamentally changing what product management means. The comfortable, process-driven PM role is dying. What's emerging in its place demands a completely different skill set, mindset, and level of technical depth.

The Three Forces Reshaping Product Management

1. Velocity Has Gone Hypersonic

When I launched my first AI feature eighteen months ago, I spent six weeks on discovery, four weeks on specs, and eight weeks on development. Last month, I went from idea to production in nine days.

This isn't an exaggeration. AI tools have compressed timelines so dramatically that traditional PM processes have become bottlenecks rather than value-adds.

Consider what's changed:

The PM who spends two weeks crafting the perfect PRD is now the PM who's holding up the team. Speed of learning has become more valuable than depth of planning.

PMs who built their careers on being the "organized coordinator" are finding their core value proposition evaporating. The half in trouble? They're the ones still scheduling meetings to discuss what could be prototyped and tested before lunch.

2. Technical Depth Is No Longer Optional

I used to believe the "PMs don't need to code" mantra. I was wrong.

You don't need to be a senior engineer, but you absolutely need to understand:

Why? Because product decisions are now deeply technical decisions.

Should you use GPT-4 or Claude? That's not an engineering question—it's a product question with cost, latency, and capability trade-offs. How do you handle hallucinations? That's a UX problem that requires understanding temperature settings and confidence thresholds.

The PMs who survive are those who can have substantive technical conversations with engineers, not just translate between "business" and "technical" teams. That translation layer is being automated away.

Singhal's right: if your primary value is writing user stories and running standups, you're in trouble.

3. The Nature of Product Strategy Has Transformed

Here's what keeps me up at night: traditional competitive moats are eroding at unprecedented speed.

A feature that would have taken your competitor six months to build now takes six weeks. Your carefully designed user experience? An AI wrapper startup just replicated 80% of it in a weekend hackathon.

This changes everything about product strategy:

Old playbook: Build differentiated features → Create switching costs → Defend your moat → Optimize incrementally

New reality: Assume any feature can be copied instantly → Compete on speed of learning → Build network effects and data moats → Reinvent continuously

The PMs in trouble are those still thinking in terms of feature roadmaps and quarterly planning cycles. The ones thriving are running continuous experimentation loops, making weekly strategic pivots based on data, and treating their product like a living organism rather than a construction project.

The New PM Archetype: Builder-Strategist

So who survives? What does the AI-era PM actually look like?

After working with dozens of teams navigating this transition, I've identified three characteristics that separate the thriving PMs from the struggling ones:

1. Hands-On Builders, Not Just Coordinators

The best PMs I know now spend 30-40% of their time actually building.

They're prototyping in Cursor or Replit. They're testing prompts. They're building internal tools. They're running their own SQL queries and analyzing data without waiting for a data scientist.

This isn't about becoming an engineer—it's about collapsing the feedback loop. When you can test your own hypotheses, you learn 10x faster than when you're dependent on others to execute your ideas.

One PM I work with built a custom AI agent to monitor user feedback across twelve channels and surface patterns. It took her four hours. Previously, she would have written a ticket, waited two sprints, and received a dashboard that was already outdated.

The ability to build gives you speed. Speed gives you learning. Learning gives you better judgment. Better judgment is what makes great PMs.

2. Comfort with Chaos and Ambiguity

The PMs struggling most are those who need clear frameworks and established best practices.

Bad news: there are no best practices for AI product management yet. We're all figuring this out in real-time.

I've watched PMs paralyzed by questions like:

The thriving PMs don't wait for answers—they run experiments. They're comfortable saying "I don't know, let's test it" fifty times a day. They treat uncertainty as the default state rather than a problem to be solved.

This requires a fundamental personality shift for many PMs. The ones in trouble are those who built their careers on being the "person with the plan." The plan changes daily now.

3. Systems Thinking Over Feature Thinking

The most profound shift I've observed: great AI-era PMs think in systems, not features.

Traditional PM: "We need to add a recommendation engine to increase engagement."

AI-era PM: "How do we create a feedback loop where user interactions improve model performance, which improves user experience, which generates more interactions?"

This is the difference between building a product and building a learning system.

Every AI product has multiple interconnected loops:

The PMs who understand these dynamics and can design products that improve themselves over time are the ones creating defensible value. The ones just shipping AI features? They're commoditizing themselves.

The Brutal Reality: Fewer PMs, Higher Bar

Here's what nobody wants to say out loud: we probably need fewer product managers overall.

When velocity increases 10x and technical depth becomes mandatory, you don't need the same PM-to-engineer ratio. A highly technical, fast-moving PM can now effectively support 15-20 engineers instead of 5-8.

Companies are already adjusting:

This isn't a temporary correction—it's a permanent restructuring of how product work gets done.

The half in trouble? They're the ones who can't operate at this new level. Mid-tier PMs who were solid coordinators but lack technical depth or building capability. Junior PMs who haven't yet developed strong judgment. Senior PMs who've spent years in process-heavy organizations and can't adapt to chaos.

Your Survival Playbook: Four Concrete Actions

If you're a PM reading this and feeling anxious, good. Anxiety means you're paying attention. Here's what to actually do:

Action 1: Build Something This Week

Not a spec. Not a roadmap. An actual working prototype.

Use Cursor, Replit, or Claude with artifacts. Build a simple AI feature, a data analysis tool, an internal workflow automation. Doesn't matter what—just build something you can show and get feedback on.

Do this every week. In three months, you'll have 12 projects and a completely different relationship with technology.

Action 2: Go Deep on One Technical Domain

Pick one area and become genuinely knowledgeable:

Read papers. Take Andrew Ng's courses. Build projects. Have conversations with engineers where you're not just nodding along.

You don't need to know everything, but you need to know something deeply enough to have earned respect.

Action 3: Shift from Planning to Experimentation

Review your last month of work. What percentage was planning versus experimenting?

Flip the ratio. Spend 30% of your time planning and 70% running experiments. Ship more half-baked ideas to get feedback. Build prototypes before writing specs. Test assumptions before debating them in meetings.

This feels uncomfortable if you've built your identity around being "strategic" and "thoughtful." Do it anyway.

Action 4: Cultivate Technical Credibility

Start contributing to technical discussions in ways you haven't before:

Your goal isn't to become an engineer—it's to become a technical product person who engineers want to work with because you make their job easier, not harder.

The Opportunity Hidden in the Chaos

Here's the paradox: while half of PMs are in trouble, the other half has never had more opportunity.

The PMs who adapt to this new reality will be more valuable, more impactful, and more compensated than ever before. Why? Because the problems are harder, the pace is faster, and the stakes are higher.

Building AI products isn't just about adding features—it's about navigating fundamental questions about human-AI interaction, trust, safety, and value creation. The PMs who can operate at this level are rare and becoming rarer as the bar rises.

I'm betting my career on this: the next generation of legendary PMs won't be the ones who perfected the old playbook. They'll be the ones who threw it out and built something new.

Singhal's warning isn't meant to scare you—it's meant to wake you up. The question isn't whether product management is changing. It's whether you're changing with it.

The gap between the PMs who thrive and those who become obsolete is widening every day. Which side of that gap are you on? More importantly, what are you doing about it?

The AI era doesn't need more coordinators. It needs builders who can think strategically, move fast, and navigate chaos. It needs PMs who can hold technical depth and user empathy simultaneously. It needs leaders who can make decisions with incomplete information and adapt when they're wrong.

If that sounds like you—or like who you're becoming—you're not in trouble. You're exactly where you need to be.

The other half? They've got work to do.