The End of the Product Manager as an Assistant: How AI Impacted Product Roles

Recruiting the right Product Management talent in 2026 is exponentially harder than it was three years ago. The candidate pool has exploded, but the signal-to-noise ratio has plummeted. It has never been easier for less qualified candidates to flood companies with AI Slop: highly polished, artificially generated applications that look great on paper but completely mask a lack of underlying substance.

At Product People, we run roughly 150 client missions a year, deploying our 50 in-house Product Management Consultants to publicly listed giants like eBay and Zalando, hyper-growth scaleups like DeepL, and complex post-M&A environments. Because we operate with a sub-0.5% acceptance rate, we’ve had to aggressively adapt to this new reality. The traditional proxies for talent (prestigious brand names, specific elite universities, and flawless take-home assignments) are no longer reliable indicators of a PM’s ability to actually deliver outcomes.

‍

The Death of Safe Bets and the Rise of Live Simulations

When the market gets noisy, the default leadership instinct is to retreat to safe bets. Companies restrict their pipelines to candidates who have worked at tier-one tech brands or direct competitors, hoping the logo on the resume guarantees competence.

I strongly disagree with this approach. Hiring exclusively from direct competitors often just leads companies to copy each other’s mistakes. We have never found traditional pedigrees to be strong indicators of success within our firm. Instead, we heavily value genuine candidate referrals and have completely restructured how we evaluate potential hires.

We quickly realized that asynchronous processing (evaluating work a candidate completed off-camera over several days) was breaking down. To combat the influx of AI slop, we toned down our rigorous three-part take-home assignment into a single, highly focused piece and transformed the traditional presentation into a dynamic, live simulation.

We need to see how people think on their feet. We need to measure their curiosity and adaptability in real-time. Recently, these live simulations have been incredibly revealing. We’ve watched candidates get stuck very early in a case, exposing a shallow, superficial thought process that a beautifully formatted CV had managed to hide. Conversely, we’ve been deeply impressed by candidates who brought a Claude Code project into the simulation, loaded it with relevant Product People context, and used it to demonstrate exceptional judgment and first-principles thinking.

We are not dogmatic about which LLM people use. Foundational model capabilities ebb and flow with every release. What matters is whether the candidate uses AI as a crutch to hide their lack of depth, or as a powerful amplifier for their own strategic logic.

‍

The PRD is (Somewhat) Obsolete

This shift in hiring directly mirrors the massive shift in the day-to-day reality of Product Management. There is a deeply ingrained industry best practice that fundamentally overvalues a PM’s ability to write User Stories and Product Requirements Documents (PRDs) from scratch.

Historically, this made perfect sense. For years, the central bottleneck in product development was translating fuzzy, high-level business thinking into precise, actionable written requirements for engineering teams. But today, that translation layer is essentially commoditized. Clear and concise documentation remains incredibly helpful, but it is infinitely easier and faster to generate with LLMs.

If you are still hiring Product Managers primarily for their ability to sit in a corner and type out lengthy PRDs from scratch, you are optimizing for yesterday's bottleneck. We can all make pretty decks and extensive documents, but at the end of the day, we prefer to deliver outcomes.

‍

The New Bottleneck: What Actually Matters Now

If translating fuzzy ideas into written documents is no longer the chokepoint, what is? The role hasn't disappeared; the expectations have escalated. The core, irreplaceable skills we now relentlessly test for during live simulations to ensure our team can drive those 150+ client missions come down to five strategic pillars:

1. Problem Selection: Figuring out which problem is actually worth solving in the first place. AI can help you build faster, but it cannot tell you if you are building the right thing.
2. Pragmatic Judgment: Demonstrating strategic judgment on what specifically drives adoption, monetization, and retention in the real world.
3. Ruthless Prioritization: This means not only ignoring the ambient noise but actively killing ideas and features when they are no longer relevant to the core strategy.
4. First-Principles Thinking: We need PMs who dissect problems down to their foundational truths rather than relying on a memorized framework or the latest hype about how "Company X" operates.
5. Moving Fast: Speed is non-negotiable. As Naval Ravikant aptly puts it: "Impatience with actions, patience with results."

‍

AI Accelerates the Build Cycle... and Product Debt

That third pillar (ruthless prioritization and the killing of ideas) is rapidly becoming the most critical survival skill in the modern product landscape.

Because AI is drastically accelerating the build cycle, engineering teams are shipping features faster than ever before. While increased velocity sounds fantastic in a board meeting, the downstream reality is that the backlog of shipped features that need to be maintained, evaluated, and potentially killed is growing exponentially. In short, AI accelerates the accumulation of product debt.

When our consultants parachute into a mission, they inevitably encounter the intense friction of corporate politics, legacy systems, and the sunk-cost fallacy. Killing a beloved feature or a pet project without alienating entrenched stakeholders is an absolute art form.

‍

The Art of the Kill: Navigating Friction

We train our teams to approach this friction methodically, starting by deeply understanding what we are actually up against. Before we pull the plug on a feature, we ask three critical questions to navigate the human and financial complexities:

First, would somebody feel this is a negative judgment on their competence? The ego is often the biggest blocker to effective product lifecycle management. We ask ourselves: How might we avoid hurting their ego? Or, even better, how can we bring them along on the journey so that sunsetting the feature feels like their own strategic decision?

Second, does killing this have a negative financial impact? We must meticulously map out both the first-order and second-order financial effects of removing the product from the market.

Third, what is the opportunity cost? What won't we be doing if we continue maintaining this legacy feature? Conversely, what level of investment would it actually take for this feature to become genuinely successful? Framing the decision as an allocation of future capital rather than a destruction of past work completely shifts the conversation.

‍

Conclusion

The AI revolution in product management isn't just about writing faster code or generating automated user stories. It is a fundamental forcing function that is stripping away the administrative busywork and exposing the true nature of product leadership.

The talent pool might be wider, and the noise might be louder, but the path forward is clear. Stop retreating to the false security of prestigious logos. Or testing for human typewriters and assistants. The Product Managers who will drive your business forward in this new era are the ones who can think from first principles in real-time, leverage AI to amplify their judgment, and possess the strategic courage to kill the things holding you back. Impatience with actions, patience with results—that is how we win.