I’m seeing one of the biggest shifts of my career in how companies operate, thanks to agentic AI. 

We’re moving beyond AI that simply provides information to AI that can actually help get work done. For chief data officers (CDOs), that shift moves the focus from the technology itself to where it can remove friction and create measurable business value. 

After more than 40 years in technology, including more than 20 in manufacturing, I know that new technologies succeed or fail based on two things: adoption and trust. 

The organizations that will win with agentic AI won’t necessarily have the most advanced tools. They’ll be the ones that help their people understand how to use those tools effectively and responsibly.

One lesson I’ve learned is that organizations rarely struggle because they don’t have enough technology. They struggle when technology, process, data, and people aren’t aligned around the same business outcome. 

Here are the agentic AI trends I’d watch, and my suggestions for what CDOs should do about them.

Trend 1: AI Is Moving From Information to Action

Agentic AI goes beyond generating content or recommendations. It can interact with systems and carry out multiple steps across a process to accomplish a goal, with varying levels of human oversight.

Agentic AI can connect information, automate routine tasks, and help employees make better decisions. Throughout manufacturing and global organizations, I’ve seen knowledge scattered across systems, documents, service records, and experienced employees. Often the challenge isn’t that the answer doesn’t exist. It’s getting the right information to the right person quickly enough to act.

Helping people find the right information, take action, and move faster is where I see tremendous potential, because it lets people spend more time solving problems and less time managing processes.

What gives me confidence is that I’ve seen this pattern before. I’ve watched technologies like the internet, mobile, cloud, and analytics move from emerging capabilities into the way businesses operate every day. Agentic AI has the potential to take that a step further by not only providing information, but helping get the work done. 

Where I think the market may be getting ahead of itself is the idea of completely autonomous enterprises. Most business leaders I talk with aren’t looking to remove people from key decisions. They’re looking to make people more effective.

In my experience, the best outcomes come when technology amplifies human judgment rather than tries to replace it. As decisions become more consequential or dependent on business context, human judgment becomes more important.

Trend 2: Agentic AI Is Moving Into Practical Use

I’m seeing the practical value of agentic AI in areas where people spend significant time gathering information, answering questions, and coordinating activities.

These include service management, employee support, cybersecurity, software development, and knowledge management. 

In manufacturing, there are opportunities in predictive maintenance, operational support, quality management, and helping frontline teams access expertise more quickly. Manufacturing has taught me that value is often created at the point where decisions are made, not where the technology sits. 

An engineer or service technician may need information from maintenance history, product documentation, quality data, and experienced colleagues. Bringing that context together quickly is where agentic AI can make a real difference.  

Speed matters, but speed alone isn’t the measure of success. I’d rather see an organization focus on fewer important business problems and deliver measurable value than launch a long list of AI initiatives that never meaningfully change the business.

Trend 3: Greater Autonomy Brings Greater Accountability

When AI starts taking action instead of simply providing recommendations, the conversation changes completely. 

If agentic AI interacts with systems or orchestrates multiple steps across a process, accountability becomes much more important.

Data and AI governance, security, auditability, and decision rights become critical. Organizations need to be clear about where human oversight is required and where automation can safely operate. 

For me, the line is based on risk and consequence. The more significant or difficult an action is to reverse, the stronger the case for human oversight. Finding information or performing a reversible task is very different from committing company funds, granting access, or making a decision with significant consequences for an employee or customer. 

I don’t see good governance as something that slows AI down. Done well, it gives an organization the confidence to move faster. Clear accountability, appropriate human checkpoints, and traceability become even more important as autonomy expands. 

The most successful organizations aren’t the ones that automate the most. They’re the ones that understand where human judgment adds value and where technology can eliminate friction.

Trend 4: From Isolated Use Cases to Enterprise-Wide Orchestration

In 2027 and 2028, what we’ll be watching most closely is the transition from isolated AI use cases to enterprise-wide orchestration. 

Many companies have already demonstrated that AI can improve individual productivity. The next question is whether it can reliably connect people, processes, and systems to create measurable business value at scale.

As we move from individual tools toward connected ecosystems of agents, the challenge grows. Business processes cross systems, functions, and decision points. That increases the importance of three things: architecture that lets agents work across systems, trusted data with clear ownership, and governance that includes monitoring and the capability for human intervention.

I’m also watching how governance and trust evolve. As AI agents gain more autonomy, organizations will need greater confidence in how decisions are made, actions are executed, and risks are managed.

Trust has to be something an organization can demonstrate, not simply assume. 

Pilots can demonstrate that something is possible. Scale is where organizations find out whether it is sustainable. 

Trend 5: The CDO Role Is Continuing to Evolve

I believe the CDO’s job description will become even more business-focused than it is today. Agentic AI will require data leaders to spend more time helping shape business strategy, workforce transformation, governance, and organizational change.

The CDO must help business leaders understand what’s possible, help technology teams understand what matters most to the business, and bring those priorities together. 

When AI is analyzing data, data quality matters. When an agent is acting on that data, quality matters even more, and so do context, access, ownership, and decision rights. The question becomes not only whether the data is available, but whether it is trustworthy enough to support an action. 

Success with agentic AI won’t belong to IT alone. It will require collaboration among business leaders, operations, HR, legal, risk, cybersecurity, and technology teams.

A CDO’s Agentic AI Action Plan

1. Start With the Business Problem

Too often, organizations focus on the capabilities of the tool rather than the outcome they’re trying to achieve. The first question should be, “What business challenge are we trying to solve, and what value will success create?”

If you can’t answer that clearly, you probably aren’t ready to talk about the technology yet. 

2. Fix the Foundation Before You Expand Autonomy

Agentic AI will expose weak data, poorly defined processes, and unclear decision rights quickly. Data quality, governance, security, and system integration may not be the most exciting topics, but they are essential to long-term success. 

I also see organizations underestimating the work required around process design and change management, which is often the difference between a successful implementation and a failed one.

Getting the foundation right doesn’t delay AI. It allows AI to scale. 

3. Teach People When and When Not to Trust AI 

People need to understand how to use AI, when to trust it, when to question it, and when human judgment needs to take over. 

Every major technology transformation I’ve experienced has required new skills, new ways of working, and strong change leadership. Technology projects are often people projects in disguise.

You can implement the technology correctly and still fall short if people don’t understand the change, trust the process, or see how it makes them more effective. 

4. Judge AI by What Changes in the Business 

It’s easy to measure how many AI tools, agents, pilots, or use cases have been deployed. I’d focus instead on what changed in the business. Did we improve quality? Reduce cycle time? Serve customers better? Increase productivity? Reduce risk? Help employees make better decisions? 

AI adoption is important, but adoption without measurable business impact isn’t transformation. 

Final Thought 

I’ve spent more than four decades watching technologies come and go. The ones that create lasting value aren’t necessarily those that generate the most excitement. They’re the ones that become part of how people work because they make the business better. 

That’s the opportunity I see with agentic AI. 

The question for CDOs isn’t how many agents we can deploy. It’s whether we can use them to make our people more capable, our decisions better, our processes simpler, and our businesses stronger.