How AI is reshaping the product role | Oji and Ezinne Udezue
A deep dive into how AI is transforming the product manager's role, featuring insights from product leaders Ezinne and Oji Udezue. They discuss essential new skills, the "shipyard" framework for development, why hands-on learning is critical, and the difference between companies succeeding and failing with AI adoption.
The Evolving Role of the Product Manager in the AI Era
The core value a Product Manager (PM) delivers—derisking product delivery and maximizing business value—remains unchanged. However, the tactical day-to-day is shifting significantly. With engineering teams accelerating their build processes, thanks to AI, PMs are freed up to invest more time in developing genuine customer insights. This acceleration also creates a new challenge: PMs can become the bottleneck if they don’t adapt.
The nature of products is also changing. We are moving from “products as artifacts,” which are static and shipped, to “products as organisms.” These are living, breathing systems with feedback loops and data metabolisms that allow them to learn and evolve. This requires a new approach to orchestration, moving beyond just managing people to managing complex technological touchpoints and data flows.
Key tactical shifts for PMs include:
- Deeper Customer Insight: Engineering teams now demand more rigorous confirmation of customer instincts and insights.
- Systems Orchestration: Managing the interplay between software, feedback loops, and LLMs to enable continuous learning within the product.
- Enhanced Data Literacy: Understanding where data comes from, how it’s organized, and how it’s leveraged for future product and customer insights is crucial.
- Creating Guardrails: A heightened responsibility to build ethical safeguards into products, a lesson learned from the social media era.
Essential Skills for PMs in 2025
To thrive in this new environment, PMs need to cultivate specific attitudes and skills.
Attitudes and Values
- Curiosity and Humility: The field is evolving so rapidly that everyone is a beginner. It’s critical to have a growth mindset, be willing to learn from anyone (including junior engineers), and admit what you don’t know. As Oji puts it, “humility is teachability, and teachability is survivability.”
- Agency and Ownership: Successful PMs act like thermostats, not thermometers—they change the temperature of the room rather than just measuring it. They see opportunities and execute without waiting for permission, demonstrating high ownership.
Hard Skills
- Beyond Prompt Engineering: A deep understanding of AI as a toolbox is necessary. This includes:
- Writing Evals: Creating robust evaluations to verify model outputs, constrain hallucinations, and ensure the AI is performing as intended.
- Multi-Model Fluency: Knowing how to use multiple models, understanding the specific strengths of each, and combining them to create a solution greater than the sum of its parts.
- Performance Tuning: Being able to tweak models through fine-tuning and weight adjustments to optimize for specific tasks.
New Frameworks for Building Products
Identifying “Sharp Problems”
The most predictive factor for success is focusing on a “sharp problem.” This means identifying a core, long-standing human need and reimagining its solution with new technology. A problem is “sharp” if a 3-10x improvement in performance or a 10x reduction in cost would make the solution irresistibly compelling. This disciplined approach helps avoid the “drunken walk” of aimless pivoting.
The “Shipyard” Model for Development
To navigate the current pace of innovation, teams should adopt a “shipyard” model that embraces controlled chaos.
- The Core Team: A six-capability pod consisting of PM, engineering, design, user research, data/ML, and product marketing. This is about ensuring the right capabilities are present, not a strict headcount.
- Constant Collaboration: The team communicates and collaborates in real-time, moving beyond the rigidity of traditional standups to solve problems as they emerge.
- Blurring Roles: Roles become less defined, requiring individuals to develop cross-functional skills.
- External Tendrils: The core team maintains strong connections—“tendrils”—to customer-facing roles like sales and support. For example, including a support manager in a design review can provide invaluable, real-world feedback and prevent shipping feature debt.
How Companies Succeed with AI Adoption
Companies that are successfully adapting to AI share common traits. They recognize AI is a core capability, not a magic pixie dust to be sprinkled on existing products.
- AI at the Core vs. AI at the Edge: The most successful companies are not just attaching AI to their existing codebase (“AI at the edge”). They are fundamentally re-architecting their solutions from a blank slate with AI as the central component (“AI at the core”). For these companies, the legacy codebase often shrinks as the LLM becomes a primary part of the problem-solving engine.
- Embrace Specificity: Instead of trying to build a single, massive LLM that does everything, successful companies build multiple, highly specialized solutions. They then create a connective tissue or a multi-model layer to tie these specific capabilities together into a cohesive offering.
- Rethink User Experience: The chat interface is not the final form of AI interaction. Companies that are winning are experimenting with novel, dynamic user experiences that personalize themselves to the customer.
The Imperative of Hands-On Learning
Theory is not enough; PMs must get hands-on with the technology. Oji notes, “I’ve written more code in the last one year than I have in the last 10 years because code is now essentially architecture and English.”
This means:
- Building Prototypes: Moving from writing static PRDs to building functional prototypes.
- Using the Tools: Subscribing to AI tools, calling APIs with Postman, and understanding different models firsthand.
- Passion Projects: The best way to learn is to solve a personal problem. Oji’s project to automate his house forces him to learn about hardware, model quantization, and fine-tuning. For PMs, the advice is clear: find something you’re passionate about and build a solution for yourself. This is the fastest path to genuine understanding.
Enduring Lessons from a 50-Year Career
- The Power of Simplicity: Complicated solutions often arise from a fear of being wrong. Great products are opinionated and simple. It takes courage to make a decision, ship a focused solution, and learn from it, rather than offering a confusing array of options. “You’re better off picking an opinion, shipping it out there, being wrong and then adjusting than leaving too many options.”
- Communication is Strategy: The best way to translate strategy into execution is relentless communication of the “why.” Every single person in the organization, down to the janitor, should understand the mission.
- Observe, Don’t Just Ask: True customer insight comes from observing what people do, not just what they say. Ethnographic research—watching users in their environment—remains the gold standard. Relying solely on AI to analyze interview transcripts is a form of cheating that misses the deeper “why” behind customer actions.
- Ethics and Responsibility: PMs wield immense power in directing technology. This comes with a profound responsibility to consider the ethical implications of what is being built.