Why Brazilian Product Managers Are Losing Relevance (and How to Get It Back)

Last week, three things happened almost simultaneously. Nikhyl Singhal, former VP of Product at Meta and Google, published a direct warning: half of all PMs are in danger. Cat Wu revealed how Anthropic’s product team works in a radically different way. And the confusion around AI tool pricing — Claude, GitHub Copilot, Cursor — exposed how drastically the cost structure of building software is changing.

These aren’t isolated events. They’re a pattern that demands attention.

What Nikhyl Singhal Is Saying (and Why It Matters)

Singhal isn’t a LinkedIn influencer making generic predictions. He’s run product at Google, Meta, and Credit Karma. When he says half of all PMs are in danger, he’s reading market patterns he’s watched play out before — in earlier waves of technology that redistributed power inside organizations.

His argument is specific: many PMs have built careers around activities that are being automated or eliminated. Writing specs. Organizing backlogs. Facilitating communication between teams. Creating product documentation.

These activities haven’t disappeared. But they no longer require a full-time person dedicated to them.

The Operational PM Trap

In Brazil, this problem is amplified. The digital product market grew fast over the last decade. Many companies hired PMs to solve one specific problem: someone needs to organize the backlog and make sure the engineering team has enough context to work.

That role is legitimate. But it’s not product strategy.

The operational PM:

• Translates stakeholder requests into tickets
• Prioritizes features based on political pressure, not impact
• Measures success by delivery, not by results
• Knows Jira well, but doesn’t know users well

When Claude Code, Cursor, or GitHub Copilot accelerate development velocity 3x or 5x, the bottleneck shifts. The problem stops being “translate decisions into code” and becomes “make the right decisions.”

And making the right decisions was never the operational PM’s job.

What Anthropic Is Doing Differently

Cat Wu, who leads product at Anthropic, recently described how the team operates. The model is different from what you see at most Brazilian companies.

PMs at Anthropic aren’t facilitators. They own vision and direction. They spend more time with users than with internal stakeholders. They define the problem before asking for solutions. And — critical point — they deeply understand the capabilities and limitations of the technology they’re building.

It’s not a model that works for every company. But it reveals a structural shift: when execution velocity increases, value moves to whoever knows what to execute.

Operational PM (at risk)

• Focused on backlog and delivery
• Prioritizes based on internal pressure
• Measures success by features shipped
• Knows management tools
• Communicates between teams

Strategic PM (in demand)

• Focused on problem and outcome
• Prioritizes based on proven impact
• Measures success by business metrics
• Knows users and technology
• Defines direction and vision

Why This Is Happening Now

Three forces converging:

AI tools accelerated execution. GitHub Copilot, Claude Code, Cursor — the time between “decide to build” and “it’s built” has shrunk dramatically. This doesn’t eliminate the need for product thinking. But it eliminates the justification for a middleman who just translates decisions into tasks.

The cost of development is in flux. Recent confusion around AI API pricing — Claude went up? Down? For whom? — exposes how drastically the economics of building software are shifting. Companies that relied on PMs to “optimize” roadmaps within engineering constraints are discovering the scarce resource isn’t engineering anymore — it’s clarity of direction.

Leadership expectations shifted. When CEOs see what small teams with AI can ship, tolerance for slow product processes drops. The question stops being “when will it be done?” and becomes “why are we building this?”

What PMs Need to Develop Now

Singhal is blunt about what separates PMs who’ll thrive from those who’ll struggle to stay relevant. It’s not technical AI skills. It’s capabilities that always separated strong PMs — but are now mandatory, not differentiating.

• Can you defend a product direction without complete data?
• Do you spend more time with users than in internal meetings?
• Can you explain why not building something matters more than your roadmap?
• Do you deeply understand the technical capabilities of what you’re building?
• Do you measure success by behavior change, not by deliverables?

If you answered “no” to more than two, Singhal’s warning applies to you.

The Skill Nobody’s Talking About

Most of the “future of PM” debate focuses on learning to use AI tools. That’s necessary, but not sufficient.

The skill that’ll separate PMs in the next few years is the ability to make judgments under uncertainty.

When you can test faster, the question shifts from “how do I validate?” to “what’s worth testing?” When you can build faster, the question shifts from “how do I prioritize?” to “what shouldn’t exist?”

PMs who thrive in this environment are the ones who can make decisions with incomplete information, defend those decisions with solid reasoning, and adjust quickly when they’re wrong — without needing six months of data to act.

What to Do in Practice

If you’re a PM in Brazil reading this with some discomfort, here’s what I’d recommend:

First, audit your time. Last week, how much time did you spend on activities AI could do (write specs, organize tasks, create documentation) versus activities requiring human judgment (interview users, make trade-off decisions, define direction)?

If the split is 70/30 toward the automatable side, you have a positioning problem, not a skill problem.

Second, increase direct user contact. Not delegated research. Not surveys analyzed by another team. Direct, regular conversation with people using (or not using) what you’re building. This isn’t product romanticism. It’s the only way to develop intuition that can’t be automated.

Third, actually learn the technology. Not to become a developer. To know what’s possible, what’s easy, what’s hard, and what’s impossible with your current stack. PMs who didn’t understand technology were always a problem. Now it’s unsustainable.

Fourth, practice making explicit decisions. Many PMs hide behind process: “we’ll prioritize in planning,” “we’ll see what the data says,” “we’ll align with stakeholders.” That’s avoiding decisions, not making them. The strategic PM says: “we’re doing X because Y, and not doing Z because W.” And owns it.

The Brazilian Paradox

Brazil’s product market has an unusual characteristic. Many companies still operate with 2018-era product structures — large teams, slow processes, PMs as backlog coordinators — while competing against startups and global companies that already reorganized how they work.

This creates an interesting window. PMs who reposition now will find demand. Companies waking up to this shift need people who work this way. And there aren’t many available.

But the window won’t stay open long. Singhal’s warning isn’t a prediction about distant futures. It’s a diagnosis of the present.

The Choice You Need to Make

There’s no neutral position here. If you’re a PM continuing to do the same work the same way, you’re choosing the risk side. Not from incompetence — from inertia.

The good news: the shift doesn’t require you to learn to code or become an AI specialist. It requires you to do the work that should always have been the PM’s job. Only now you can’t put it off anymore.