AI isn't failing because the technology is lacking. It's failing because no one is truly taking responsibility for the results. While everyone debates whether Product Managers are still needed, the reality is that AI projects are failing almost twice as often as regular software projects. It's a tough pill to swallow, especially when every headline seems to say that AI will take over our jobs. But this isn't about bad code. It's about missing the basics of good product management. As someone who's spent years in product leadership, I keep seeing the same story play out again and again:
AI will replace Product Managers.
After more than ten years leading product teams in tough, highly regulated industries, I see things differently.
AI is not exposing the irrelevance of product management. It is exposing the absence of it.
This article explores the "Airbnb Myth," the reality of AI ROI timelines, and the specialized blueprint required for product leaders to bridge the gap between technical capability and human value.
Part I: The Ghost in the Machine
The Data We Keep Ignoring
Let's start with reality.
In 2025 alone:
- 42% of companies abandoned most of their AI initiatives, up from just 17% the year before
- 46% of AI proofs-of-concept never reached production
- Only 6% of organizations reported achieving AI ROI within 12 months.
At the same time, LinkedIn is flooded with confident predictions about the end of product management.
- AI will handle discovery.
- AI will prioritize roadmaps.
- AI will decide what to build next.
And companies like Airbnb are held up as proof that PMs are no longer needed.
Here is the contradiction we rarely examine.
The same organizations declaring product management obsolete are often the same ones quietly watching their AI initiatives collapse.
That is not a coincidence.
The Great AI Disconnect
We are in the midst of a major technological hype cycle, with many claiming AI will automate the Product Manager role out of existence.
But here is the uncomfortable truth, the "AI will do everything" crowd doesn't want to discuss: Most AI projects are currently setting money on fire.
The average organization is scrapping nearly half of its AI proofs-of-concept before they ever reach production.
If AI were the self-managing, self-correcting force the futurists claim, why are the failure rates so high? Because AI is a calculator, not a super-intelligence. And a calculator without a mathematician is just a box of parts.
The Airbnb Narrative That Refuses to Die
We cannot discuss the "Death of the PM" without addressing the Airbnb narrative. When Brian Chesky announced that Airbnb "got rid of the classic product management function," the industry erupted.
Headlines declared PM dead.
Designers celebrated.
Think pieces multiplied.
But what happened next rarely gets shared.
Chesky later clarified:
I should have been clearer. We morphed the function into an Apple-style product marketing function.
He went even further:
Make no mistake, product managers are critical. They just shouldn't be doing the job of a designer.
As an executive who has navigated these cycles for over a decade, I saw something different: Evolution.
Airbnb didn't fire their PMs; they elevated them. They realized that in an AI-driven world, the "generalist" PM who just manages Jira tickets is indeed obsolete. What survived, and what Airbnb prioritized, was the Strategic PM. The leader who can tell a story, define a go-to-market strategy, and ensure the product resonates with human needs.
Part II: The Real Diagnosis
Why AI Projects Fail (The Data)
When we look at the wreckage of failed AI initiatives, the root causes are rarely technical. Research from the RAND Corporation identifies three primary points of failure:
- The Problem Misalignment: Stakeholders don't actually know what problem they are trying to solve.
- The Data Void: Organizations lack the clean, structured data required to train effective models.
- The "Shiny Object" Trap: Companies prioritize using the latest LLM over solving a user pain point.
According to Informatica's 2025 CDO survey, the top obstacles are data quality (43%), lack of technical maturity (43%), and a shortage of skills (35%).
Notice the pattern: These are Product Problems.
No organization is failing because they have "too many PMs" asking about user value. They are failing because they launched without the discipline that Product Management provides: specificity, accountability, and a relentless focus on usability.
The Timeline Reality Check
I've watched this play out across healthcare and finance. Teams get caught in the "Magic Calculator" trap. They treat AI like a supernatural force that will deliver ROI in 90 days.
Deloitte's research tells a different story. Most organizations achieve satisfactory ROI on AI projects within two to four years. Only 6% reported payback in under a year.
That 12-to-24 month window isn't a failure; it's the Calibration Period. This is where the PM becomes the "Missing Piece." The PM is the one who sets these expectations, builds the quality gates, and manages the stakeholders through the "trough of disillusionment."
Part III: The Market Reality
Normalization vs. Obsolescence
The "PM is dead" narrative falls apart when you look at the job market. There are over 6,000 open PM roles globally right now - 53.6% above the bottom of 2023.
Specifically, AI PM roles have surged. Today, there are nearly 700 open positions at companies like OpenAI, Anthropic, and Salesforce.
We aren't seeing the death of a role; we are seeing the birth of a specialty.
- Finance matured.
- Engineering matured.
- Now, Product Management is maturing.
General-purpose roles are decreasing because the market no longer needs "general" help. It needs domain-specific experts who understand how to integrate AI into regulated workflows like healthcare and finance without breaking the user experience.
Part IV: The PM Advantage in the AI Era
The Great Misunderstanding About Product Management
Before we start discussing how Product Management can solve these issues, we have to solve the common misunderstanding about what Product Management really is.
If product management were simply:
- Writing PRDs
- Updating roadmaps
- Running ceremonies
- Managing backlogs
Then yes, automation would replace much of it.
But that is not the job. Product management is a vocation.
- It is accountability.
- It is judgment.
- It is being the person responsible for translating human problems into measurable outcomes and owning the result when reality does not match the plan.
AI can generate options. It cannot carry accountability. And accountability is exactly what most AI initiatives lack.
The Translation Layer
Product management is the translation layer between what technology can do and what humans need to do.
MIT's research confirms that how companies adopt AI is the difference between success and bankruptcy. Internal builds succeed only one-third of the time. Partnerships and specialized vendor integrations succeed 67% of the time.
The difference-maker? A Product Leader who asks:
- "Does this solve a core user friction point?"
- "Is our data ready for this?"
- "What happens when the model hallucinates in a regulated environment?"
If a system is accurate but painful to use, it is a failure. I have seen "perfect" models killed by poor UX more often than by technical bugs.
Part V: The Framework for AI Success
For those leading AI integration, I propose a Product-First Blueprint:
1. Enforce Ruthless Specificity
Stop trying to "AI-enable" your entire platform. Organizations that see financial returns are twice as likely to have redesigned a single, end-to-end workflow before even selecting a model. Solve one problem completely.
2. Set "Executive" Timelines
Stop promising 90-day magic. Communicate that value realization takes 13+ months. Building this into stakeholder expectations from day one is what separates success from failure. Trust is built through honesty, not optimism.
3. Resource adequately
AI initiatives starved of data, budget, or talent rarely recover. 95% of AI ROI leaders allocate more than 10% of their technology budget specifically to AI. Don't put the cart before the horse.
4. Usability Over Accuracy
A 90% accurate model that fits into a user's flow beats a 99% accurate model that requires a manual. UX is the only metric that scales, and adoption beats perfection every time.
5. Focused Feedback Loops
Small, engaged user cohorts generate better insights than "Big Bang" releases. Use high-touch validation groups to find the "hallucination edge cases" before you scale.
Conclusion: The Future of Accountability
The 80% failure rate isn't evidence that AI is overhyped. It is evident that the discipline of Product matters more than ever.
As we navigate this transition, remember that market normalization and role specialization aren't the death of our profession; they're its natural evolution. The question isn't whether AI will replace Product Managers, but how the best Product Managers will harness AI to create unprecedented value. AI does not eliminate the need for product management. It clarifies it. Because someone still has to own outcomes. And until that ownership is explicit, AI will continue to fail loudly.
If your organization is struggling with AI adoption, the answer isn't fewer PMs. It's better to have PM engagement earlier in the process.
Who in your organization is positioned to fix the focus, the timelines, and the UX? And are you letting them?