An interview with Chika Nkwocha, Senior Product Officer at Ministry of Housing Communities and Local Government (MHCLG).. Chika has led multiple AI agent implementations across enterprise products and is known for his thought leadership on the evolving role of product management in the age of AI.
Interviewer: Thank you for joining us today, Chika. Let’s dive right in. How are AI agents fundamentally changing the product manager’s role?
Chika Nkwocha: The transformation is profound. Traditional product management centered around feature roadmaps and user stories. Now, with AI agents, we’re orchestrating capabilities and experiences that can evolve autonomously. Product managers are becoming less like architects with blueprints and more like gardeners who create the conditions for intelligence to flourish. We set the boundaries, provide the training data, and establish the guardrails—but the specific behaviors emerge rather than being explicitly programmed.
Interviewer: That’s a fascinating metaphor. How does this change the day-to-day work?
Chika Nkwocha: The shift is dramatic in three areas. First, requirements definition has evolved from “what should this button do?” to “what should this agent understand and prioritize?” Second, testing has become probabilistic rather than deterministic—we’re evaluating distributions of behavior rather than fixed functions. And third, our relationship with engineering has transformed. Engineers aren’t just implementing specs; they’re co-creating intelligence architecture with us.
Interviewer: Given these changes, what new skills do product managers need to develop?
Chika Nkwocha: Statistical thinking is non-negotiable now. You need to understand confidence intervals, error rates, and performance distributions. Prompt engineering—or more broadly, agent instruction design—has become a critical skill. But perhaps most importantly, PMs need sophisticated ethical frameworks. When your product can generate novel content or make autonomous decisions, you must have robust frameworks for evaluating emergent behaviors against your company’s values.
Interviewer: How has the relationship with stakeholders changed?
Chika Nkwocha: It’s become both easier and harder. Easier because AI agents can rapidly adapt to new requirements without complete redevelopment. Harder because managing expectations about AI capabilities requires constant education. I’ve found myself spending much more time helping executives understand what’s technically feasible versus what’s still science fiction. The most successful product leaders now are translators between business objectives and AI realities.
Interviewer: Let’s talk about metrics. How do you measure success differently with AI agents?
Chika Nkwocha: We’ve had to develop entirely new measurement frameworks. Beyond traditional metrics like task completion, we now track things like agent autonomy rate; how often the system can complete tasks without human intervention. We measure confidence calibration—whether the agent knows what it doesn’t know. And perhaps most importantly, we track user trust accumulation over time. The best AI products build trust gradually by demonstrating increasing competence while remaining transparent about limitations.
Interviewer: Many products now include both human and AI components. How do you manage these hybrid systems?
Chika Nkwocha: This is where product management becomes truly fascinating. We’re designing not just interfaces but relationships. The key is finding the right level of agency for each component. In some cases, the AI should defer to humans. In others, it should act independently but inform humans. And in still others, it should handle everything unless it encounters exceptions. These agency levels need to be fluid and context-sensitive. The art is in designing when and how these transitions occur.
Interviewer: How has roadmap planning changed with AI agents?
Chika Nkwocha: It’s shifted from feature-based to capability-based planning. Instead of listing discrete features, we map out intelligence thresholds. For example, rather than saying “add sorting to the dashboard,” we might define “develop contextual awareness of user priorities.” This requires much more flexibility in planning. We set directional goals for the intelligence we want to build, but we need to be adaptive about the specific manifestations as we learn what’s possible.
Interviewer: What impact has this had on team composition?
Chika Nkwocha: Product teams now need embedded AI ethicists, data scientists, and prompt engineers alongside the traditional roles. Product managers must be comfortable directing these specialists without necessarily having their deep technical knowledge. It’s become much more like film directing, where you guide experts from different disciplines toward a unified vision without necessarily knowing how to operate every piece of equipment yourself.
Interviewer: Are there new processes you’ve found effective for managing AI agent products?
Chika Nkwocha: Absolutely. We’ve implemented what we call “emergence reviews” where we regularly examine unexpected behaviors and assess whether they’re beneficial serendipity or problematic edge cases. We’ve also established “capability governance boards” that evaluate whether new agent abilities should be released based on their potential impacts. And we run extensive red-team exercises where we stress-test agents against adversarial inputs. These processes are now as fundamental as sprint planning once was.
Interviewer: What advice would you give to product managers transitioning into AI-first roles?
Chika Nkwocha: First, invest in understanding the technology enough to have intelligent conversations with specialists. You don’t need to be an ML expert, but you should understand the capabilities and limitations of different approaches. Second, develop a strong ethical framework. The hardest questions in AI product management aren’t technical—they’re about values and tradeoffs. And finally, embrace uncertainty. The products that will win aren’t those with the most features but those that create the most adaptable, trustworthy intelligence.
Interviewer: Looking ahead, what do you see as the next frontier for product management in AI?
Chika Nkwocha: I believe we’re moving toward what I call “emergent product management,” where PMs define desired emergent properties rather than specific behaviors. For example, instead of defining how an agent should respond to specific inputs, we’ll establish broader qualities like “maintains consistency across interactions” or “demonstrates appropriate caution in high-stakes domains.” This requires a totally different mindset—moving from instructing to influencing, from controlling to cultivating.
Interviewer: That’s a profound shift. Final question: What excites you most about this transformation?
Chika Nkwocha: The opportunity to create products that truly understand and adapt to human needs in ways that were previously impossible. At their best, AI agents can reduce friction, anticipate needs, and create experiences that feel magical. As product leaders, we’re no longer just building tools—we’re creating intelligent companions that amplify human capabilities. That’s an extraordinary privilege and responsibility. The product managers who embrace this shift will define the next generation of truly transformative products.
Interviewer: Thank you for these insights, Chika. You’ve given us much to think about regarding the future of product management in an AI-first world.
Chika Nkwocha: My pleasure. It’s an exciting time to be in this field; challenging, certainly, but filled with potential to fundamentally improve how people interact with technology.
