I Didn’t Just Learn to Use AI. I Learned to Govern It.

Like many people, I began by asking AI questions.

Then I started using it to help me think through problems, organize information, develop content, conduct research, improve workflows, and explore ideas.

Before long, I was not simply asking, “What can AI do?” I was asking, “What can I do differently now that this capability exists?”

That distinction changed everything.

Eventually, using someone else’s AI tools was not enough. I wanted to build.

I began thinking about websites, applications, knowledge systems, AI agents, workflows, integrations, and automation.

I started thinking about how information moves from one system to another. What should trigger the next action? What information should an AI system receive? What happens after it produces an answer? Could an AI agent perform part of the work? Could several systems work together?

Could something I had done manually for years become faster without sacrificing quality?

Those questions moved me from user to builder. And building introduced an entirely different set of questions.

Once you build systems, you eventually have to operate them.

That is when the shiny-object phase of AI starts wearing off.

It is easy to connect another tool. It is harder to decide whether that tool belongs in your technology stack at all.

It is easy to automate something. It is harder to determine whether it should be automated.

It is easy to generate information. It is harder to determine whether that information is accurate enough, sufficiently sourced, and appropriate for professional reliance.

And when you work in professions involving confidential information, fiduciary responsibilities, legal obligations, and people’s property, businesses, families, and legacies, those questions are not theoretical. They matter.

I did not set out to become an AI governance educator.

In many ways, governance found me through the process of becoming a more sophisticated AI user.

The more capable these systems became, the more I found myself asking questions that had very little to do with prompts:

What information should this AI be allowed to access? Where is that information stored? Who can see it? What happens to confidential information? When should information remain local rather than being sent to a cloud model? What actions can an AI agent take independently? Which actions require human approval? Can we determine what information produced a particular result? Can we reconstruct what happened if something goes wrong? Who remains accountable?

Those are not merely AI questions. They are governance questions.

And that realization gave me language for my own evolution.

There will always be another model. Another application. Another agent. Another integration. Another company promising that its technology will automate everything.

I do not need everything automated.

I need the right things intelligently automated within the right structure.

My next stage is therefore not about collecting more AI tools. It is about building better governed systems.

It means deciding intentionally what AI should know. It means controlling what AI is permitted to do. It means preserving human judgment where judgment matters. It means protecting confidential and sensitive information. It means maintaining provenance so that important information can be traced back to its source. It means creating approval gates for consequential actions.

And it means asking a question that sometimes gets lost in all the excitement: Is this technology actually making the work better?

One of the things I have come to believe most strongly is that AI does not diminish the importance of expertise.

It increases the importance of knowing when expertise is required.

An AI system may help draft a legal document. That does not make the AI a lawyer.

Someone still has to determine whether the document accurately reflects the law, the client’s circumstances, the applicable jurisdiction, the client’s objectives, and the professional obligations surrounding the representation.

The same principle applies across professions. Real estate. Accounting. Financial services. Healthcare. Engineering. Education. Consulting.

AI can accelerate the work of experts. It can expand what experts are capable of producing. It can help professionals organize decades of accumulated knowledge and deploy that knowledge more efficiently.

But speed does not eliminate responsibility.

And efficiency without governance can create liability just as quickly as it creates productivity.

Perhaps more than when I started.

I still get excited when I discover something AI can do that I could not do yesterday.

I still experiment. I still build. I still have those moments when I look at something we have created and think, “This is extraordinary.”

But my relationship with AI has matured.

In the beginning, I wanted to know what AI could do. Then I wanted to know what I could build with it. Now I am equally interested in another question: What should we build, and how should we govern it once we do?

That may be one of the clearest signs of maturity with powerful technology.

Capability is not the only question.

The question is what we do with the capability.

AI can assist expertise. It can amplify expertise. It can make expertise more efficient, accessible, and scalable.

But experts still have to drive.

And I increasingly believe that the professionals who derive the greatest long-term value from artificial intelligence will not simply be the people who learn the most prompts or accumulate the most tools.

They will be the people who learn how to govern what they build.