How AI is reshaping the product role | Oji and Ezinne Udezue

When product management meets a world that only gets weirder

The most consequential shift in product work today isn’t a single tool or a single model. It is a change in tempo and in role: software stops being a static artifact and begins to behave like an organism, constantly breathing in data and exhaling decisions. That change remakes what success looks like for product leaders, and recalibrates the skills, team shapes, and questions that matter.

From artifact to organism: the new metabolism of products

Products once shipped and waited; now they learn. Behind that modest sentence sits a cascade of implications. Instrumentation becomes a core practice rather than an afterthought. Feedback loops—human and machine—become the product’s metabolism. Leaders who treat their software like living systems rewire roadmaps around continuous learning, not one-off releases.

Data literacy as a baseline

When the product is feeding on user behavior and model outputs, everyone on the product team needs fluency in where data flows and how it’s used. That means establishing clear telemetry, designing experiments to generate reliable signals, and documenting how datasets evolve. Those practices let teams distinguish between a trend and an artifact of a hallucinating model.

The pragmatic model: shipyards and controlled chaos

Controlled chaos is not an invitation to disorder; it’s a prescription for skilled orchestration. A shipyard—a compact, cross-functional unit composed of six capabilities—becomes the organizational unit best suited for rapid iteration in an AI era. That shipyard includes product, engineering, design, user research, data/ML expertise, and product marketing as a tight pod, communicating hourly rather than once a day.

Why six capabilities, not fixed roles

The emphasis is not on headcount but on capability. Shipyards borrow from industrial metaphors: what looks like frenetic motion is actually coordinated choreography of specialists and tendrils that reach into sales, support, and customers. The pod model is designed to make decisions faster, validate hypotheses more quickly, and adapt user experience dynamically.

Pick a sharp problem, not a shiny answer

One of the clearest principles for long-term product success is sharp problem selection. A sharp problem is frequent, painful, and susceptible to dramatic improvement—think 3x to 10x better or far cheaper. Reimagining an old, stubborn pain with new technology is often more defensible than inventing an entirely new need. That clarity of focus helps teams resist chasing novelty and instead design for real, measurable value.

Simplicity and conviction

Successful teams are opinionated. They pick a design path, optimize it, and measure whether users actually adopt it. Too many choices create murky experiences that obscure learning. The faster route to clarity is to ship an opinionated, simple experience, observe, and iterate. That concretizes learning and accelerates product-market fit.

Hands-on learning: code, prototypes, and a smart toaster

Learning the craft of modern product work requires more than briefings. The most effective practitioners are writing code, building prototypes, and running experiments. That hands-on approach is both practical and strategic: prototypes clarify the architecture, make trade-offs tangible, and reveal where models succeed and fail.

One practical pattern is to choose a personal project that forces you to use multiple tools—APIs, on-device models, fine-tuning, and edge inference. These projects shift theory into muscle memory; they teach how to compose models, how to constrain hallucination, and how to instrument behavior in the wild.

New hiring criteria for the AI age

Recruiting product talent now privileges a mixture of dispositions and technical practices. Curiosity and humility top the list—candidates must be teachable in a field where blueprints are still being written. Agency and ownership are equally important: teams want people who will act like thermostats, not thermometers, changing the temperature of the room rather than only reporting it.

Practical skills include data literacy, the ability to create meaningful evals for model outputs, and familiarity with multi-model orchestration. Teams that intentionally hire for these behaviors avoid the trap of looking for embedded apostles of a single tool and instead prioritize adaptable problem solvers.

AI at the core vs. AI at the edge

Not all AI integrations are equal. Many companies will add models at interfaces—an assistant here, an autocomplete there—and call that AI. The more disruptive route is to reimagine the core workflow with models as the primary mechanism of intelligence. Successful companies either design new products grounded in models or carefully migrate legacy codebases toward model-first architectures while preserving revenue during the transition.

Ethics, responsibility, and the social dimension

With great product power comes structural responsibility. Ethics cannot be an afterthought or checkbox; it must be an operational concern with guardrails, audits, and a culture that questions the implications of the software being shipped. Product teams are now builders of societal tools, and the decisions embedded in design influence behaviors at scale.

Career design: teachability is survivability

At the individual level, the dominant career heuristic is teachability. For product people, humility is shorthand for being willing to learn practical skills, even from juniors. The most resilient careers are built by people who are willing to relearn, prototype, break things, and iterate their own capabilities alongside the technology.

Final thought

The future of product work is not a single revolution but a long reorientation: toward living systems, cross-disciplinary teams, and relentless attention to what users actually do rather than what they say. That orientation favors people who are curious, humble, and hands-on—leaders who can translate chaotic motion into coordinated progress and who treat technology not as magic but as a set of capabilities to be responsibly wielded. The work ahead is less about outrunning change than about learning to live with it, and shaping it in ways that actually make life easier.