The Rise Of AI-Powered Product Governance: From Institutional Knowledge To Scalable Decision-Making

Artificial intelligence has evolved from an experimental technology into a core component of modern business operations. Yet for most organisations, the biggest challenge is no longer adopting AI, it is scaling it and embedding it into everyday business processes. According to Deloitte’s latest State of AI in the Enterprise report, employee access to AI tools has increased by 50% over the past year. At the same time, business leaders increasingly identify governance, scalability, return on investment, and the safe integration of AI into enterprise workflows not the technology itself as their primary challenges.

One area where this shift is becoming particularly visible is product governance. As digital products grow more complex, organisations must ensure they meet internal standards, regulatory requirements, security policies, and customer expectations without slowing down innovation.

Traditional governance models, built around documentation, expert committees, and manual reviews, are reaching their limits. AI-powered systems are emerging as a way to structure institutional knowledge, streamline review processes, and support faster, more consistent decision-making.

Anastasiia Davii is a Product Manager at Visa, where she led the development of an AI-powered system for automating product compliance reviews for global payment solutions. With over a decade of experience in banking and payments, she previously helped build one of Europe’s largest digital identity platforms, used by more than 60 million people. Her work focuses on one of enterprise AI’s biggest challenges: turning institutional knowledge into intelligent systems that support faster, more consistent decision-making at scale.

Today, Most Organisations Use AI To Improve Individual Productivity. Your Work Focuses On A Very Different Challenge; Bringing AI Into Product Governance. What Makes Product Governance Such A Compelling Area For AI?

Product governance is one of the first enterprise processes where AI has the potential to improve the quality of decisions, not just the speed of execution. It sits at the intersection of customer experience, engineering, compliance, security, and business priorities, making consistent decision-making increasingly difficult as products become more complex.

The first wave of AI focused on individual productivity. The next is about organisational decision-making. That’s where I see AI creating the greatest value, not by replacing human judgment, but by making organisational knowledge available exactly when decisions are being made.

When Organisations Try To “Put Their Knowledge Into AI” They Often Underestimate How Difficult That Actually Is. What Makes Transforming Scattered Organisational Knowledge Into Something AI Can Reliably Reason Over So Challenging?

Companies often assume they’re solving an AI problem, when in reality they’re uncovering a knowledge problem.

Most organisations already have the information they need. The challenge is that it’s scattered across documents, historical decisions, emails, and individual experts. Before AI can reason over that knowledge, it first has to be structured, connected, and made consistent.

The hardest part isn’t collecting information, it’s capturing context. Why was a decision made? Does it still apply? Are there exceptions? Much of that knowledge was never documented because people simply carried it in their heads.

That’s why I believe knowledge engineering is becoming just as important as AI engineering. As AI becomes easier to deploy, the real challenge is no longer building another AI assistant it’s creating a trusted knowledge foundation that every AI system can rely on.

In What Situations Does AI Still Fall Short Of Human Judgment? What Do Those Limitations Tell Us About The Real Potential And The Limits Of Scaling Expertise Through AI?

AI is exceptionally good at processing information, but judgment is a fundamentally different challenge. It can analyse large volumes of documentation, connect information from multiple sources, identify patterns, and surface insights that would take people much longer to find. That’s where AI creates the greatest value.

Where it still falls short is in situations where there isn’t a single correct answer. Product decisions often require balancing customer needs, business priorities, regulatory requirements, and long-term strategy.

Those trade-offs depend on context, experience, and business judgment, things AI can support, but not fully replace.

That’s why I see domain-specific AI as far more valuable than general-purpose models. When AI is grounded in an organisation’s own knowledge and operating principles, it can provide much more relevant recommendations. But recommendations are not decisions.

This is also consistent with research by Harvard Business School and BCG, which found that AI delivers the greatest gains when it augments human judgment rather than replaces it. The real opportunity is to scale expertise, not judgment. AI should help people make better-informed decisions, while accountability, especially in highly regulated industries remains with people.

In Financial Services, What Are The Real Business Risks When Review And Governance Processes Fail To Keep Pace With The Speed And Complexity Of Modern Digital Product Development?

Governance is often perceived as bureaucracy because people usually encounter it at the very end of the product lifecycle during legal, compliance, or risk reviews. By then, any issue feels like a delay rather than valuable feedback.

Working on an AI-powered product review system at Visa changed the way I think about governance. I realised that the biggest challenge wasn’t reviewing decisions, it was helping teams make better decisions before the review ever happened. The goal wasn’t simply to automate the final review. It was to make product requirements, compliance knowledge, and previous decisions available much earlier in the product development process.

When critical information becomes available earlier, teams can identify issues while they’re still easy to address, rather than just before launch when changes become costly and time-consuming. That’s why I believe the future of governance is shifting from reviewing decisions to supporting decisions.

When teams have access to the right knowledge while they’re making decisions not after they’ve made them, governance stops being a bottleneck and becomes a mechanism for reducing uncertainty early. That’s where I believe AI creates the greatest value.

If AI Is Responsible For Reviewing Products Against Internal Standards And Compliance Requirements, How Do You Ensure That The AI Itself Remains Transparent, Explainable And Trusted By The Teams Whose Work It Evaluates?

I would be careful with the idea that AI is responsible for compliance decisions. AI can support decisions, but accountability should always remain with people.

Working on AI-powered product reviews reinforced how important explainability is. If a system recommends that something is non-compliant, people need to understand why. That means showing which requirements the recommendation is based on, what evidence was considered, and how confident the system is in its assessment.

People also need to be able to challenge the recommendation. If an AI system can’t explain its reasoning or its conclusions can’t be verified, it becomes a black box and black boxes don’t belong in high-stakes decision-making.

I think of AI as a very knowledgeable colleague who can provide a well-informed second opinion. It can highlight risks, surface relevant information, and challenge assumptions. But the final decision and the accountability for that decision should always remain with people.

In my experience, people don’t trust AI because they’re told to. They trust it when they understand its reasoning and when it consistently helps them make better decisions. That’s why explainability and human oversight are becoming fundamental principles of enterprise AI and are increasingly reflected in regulatory frameworks such as the EU AI Act.

How Do You See The Role Of Product Managers Evolving As AI Takes On More Responsibility For Analysing Requirements, Reviewing Solutions And Generating Recommendations? What Skills Will Become Most Valuable In The Years Ahead?

AI will undoubtedly change the role of Product Managers, but not by making them less important. If anything, it will raise the bar.

I’ve already noticed this changing the way I work. Instead of spending hours gathering information from different people and documents, I now spend much more time testing assumptions, challenging recommendations, and deciding which trade-offs actually make sense.

As AI takes over more of the operational work: analysing requirements, synthesising information, or generating recommendations the differentiator will no longer be access to information. It will be the ability to exercise judgment. Product Managers will spend less time collecting inputs and more time defining problems, making decisions under uncertainty, and aligning customer needs with business priorities.

I also believe that asking the right questions will become just as important as finding the right answers. The quality of AI output ultimately depends on the quality of the problem you’re trying to solve.

The best product managers won’t be those who use AI the most, they’ll be the ones who know when to trust it, when to challenge it, and how to combine it with critical thinking, business judgment, and domain expertise.

Looking Five Years Ahead, What Do You Think Product Governance Will Look Like In Leading Technology And Financial Companies? Which Principles And Practices Do You Believe Will Become The New Industry Standard?

I don’t think the biggest change will be AI itself. The biggest change will be the way organisations treat knowledge.

Working on AI-powered product governance convinced me that knowledge is becoming infrastructure rather than documentation. Today, governance is largely document-driven. Teams interpret policies, requirements, and previous decisions manually, which becomes increasingly difficult as products grow more complex.

Over the next five years, I expect governance to become knowledge-driven. Instead of relying on static documentation, organisations will increasingly build structured knowledge systems that AI can reason over. That will make governance more consistent, easier to scale, and less dependent on individual experts.

I also don’t think long-term success will be determined by who has the best AI model. AI models will continue to improve and become more accessible. What will be much harder to replicate is an organisation’s ability to capture experience, operationalise institutional knowledge, and continuously learn from previous decisions.

The companies that succeed won’t simply use AI more effectively, they’ll learn faster because they’ve turned their institutional knowledge into a strategic asset.