Most business leaders I talk to are thinking about AI.

They are asking sensible questions. How can AI help our employees work faster? Can it analyze our data? Can it improve customer service? Can it automate some of the work people do every day?

Those are good questions. But I am increasingly convinced they are not the most important questions.

There is another one that established companies may need to start asking:

What happens when someone uses AI not to improve your business model, but to build a completely different one?

That is where AI starts to look less like another business tool and more like a potential source of disruption.

We Have Seen This Movie Before

I lived through the move from DOS to Windows and then through the rise of the Internet. In both cases, the early discussion focused heavily on what the new technology could do for the businesses that already existed.

Then something more important happened. New businesses appeared that could not have worked the same way before the technology existed.

Amazon is probably the easiest example for me to understand.

Before online shopping, buying a book was simple. You went to a bookstore. You browsed the shelves, picked up a few books, talked with someone who knew what was available, bought one, and took it home.

There was nothing obviously wrong with that model. In fact, many people enjoyed it.

Then along came Amazon.

Amazon Did Not Just Build a Better Bookstore

Imagine running a bookstore in the early days of the Internet.

You might have asked how the Internet could help your existing business. Maybe you could create a website, show customers what books were available, post store hours, or eventually allow them to reserve a book online.

Those would all have been reasonable ideas.

They also would have missed the larger change.

The more disruptive question was:

If the Internet exists, why does the store have to be the center of the business at all?

Once you ask that question, a whole series of assumptions starts to fall apart.

The customer does not necessarily need to travel to the store. The inventory does not need to fit on the shelves of a local building. The store does not need to close at night. The customer base does not need to live within driving distance.

The Internet did not simply make the bookstore more productive.

It allowed someone to rethink what retail could be.

That distinction is worth remembering as we think about AI.

Most Companies Are Still Asking the Productivity Question

Much of the business conversation around AI today is focused on productivity.

That makes sense. AI can help people write, research, summarize meetings, analyze spreadsheets, create software, answer questions, and work through large amounts of information.

I use AI that way myself.

AI can also act as an advisor. It can help a business leader think through a problem, challenge an assumption, model possibilities, or uncover something hidden in the data.

All of that matters.

But these are still mostly versions of the same question:

How can I add AI to the company I already have?

There is a much larger question sitting behind it:

What kind of company can exist now because AI exists?

Those two questions can lead to very different answers.

Meet the 22-Year-Old AI Expert

Imagine a smart 22-year-old looking at your industry.

She understands AI agents, software, APIs, data, automation, and how quickly modern systems can communicate with one another.

What she probably does not have is your 30 years of experience.

That is a real disadvantage.

She does not know your customers the way you do. She does not have your supplier relationships. She has not lived through your industry cycles. She does not understand all the strange things that can go wrong.

But she also does not have something else.

Your assumptions.

She does not know that the report has always been produced on the third business day of the month.

She does not know that customers are supposed to call the sales department for a quote.

She does not know that someone has to download information from one system, clean it up in Excel, and email it to six people.

She does not know that a salesperson is supposed to notice when an account has stopped buying.

She does not know that one department emails another department, where someone enters the same information into another system.

She looks at all of that with fresh eyes and asks a very dangerous question:

Why?

Why Does a Human Have to Do That?

This is where I think AI starts to become much more interesting.

Why does someone have to prepare the report?

Maybe nobody does. Perhaps an intelligent system watches the underlying data all the time and tells the right person when something actually needs attention.

Why does a salesperson have to remember that a customer has not reordered?

Perhaps an agent detects the change, looks at the customer’s history, investigates possible reasons, identifies the size of the opportunity, and prepares the next action.

Why does someone have to reconcile two systems every month?

Perhaps an agent watches the transactions continuously, matches the easy ones, investigates differences, and sends unusual exceptions to a person for review.

Why does a customer have to call or email the company just to get information?

Eventually, the customer’s AI may be able to communicate directly with the company’s AI.

At that point, the question has changed.

It is no longer:

How can AI make this process faster?

It becomes:

Why do we still have this process at all?

That is the beginning of disruption.

Your Most Dangerous Competitor May Not Look Like You

Established companies naturally watch established competitors.

A manufacturer watches other manufacturers. A distributor watches other distributors. A consulting firm watches other consulting firms. A software company watches other software companies.

We compare prices, products, employees, market share, service, customers, and capabilities.

But disruptive competitors often do not look very threatening at first because they are not trying to play the old game better.

They are changing the game.

The biggest AI threat to an established business may not be that its traditional competitor adopts ChatGPT more aggressively.

It may be that someone builds a new competitor around AI from the beginning.

That company may require fewer people to coordinate work. It may make decisions continuously instead of waiting for the monthly report. Its systems may not just provide information; they may take authorized action.

Its customers may interact with intelligent systems rather than working their way through a traditional organization.

Its cost structure could be different.

Its speed could be different.

Its customer experience could be different.

And those differences may compound.

This Does Not Mean People Disappear

There is an easy mistake to make when talking about this.

If AI can analyze information, communicate, and take actions, it is tempting to jump to the conclusion that companies will eventually have almost no employees.

I do not think we know that.

Businesses still need judgment. They need accountability. They need relationships, creativity, leadership, trust, and people who understand things that are difficult to put into a database.

AI systems also make mistakes.

An agent that can act needs authority limits. It needs monitoring. It needs rules about when to stop and ask a person. Someone needs to be accountable for what it does.

So the important question is not simply how many people AI can eliminate.

The more useful question is:

What should people be doing when machines can increasingly handle information, analysis, coordination, and routine action?

That could lead to a very different kind of organization.

Established Companies Have Powerful Advantages

The 22-year-old AI entrepreneur has another problem.

She may have the technology, but she does not have what an established company already possesses.

You may have decades of transaction history. You have customer relationships, supplier relationships, experienced employees, operating knowledge, physical assets, a reputation, distribution, and a place in the market.

Those are real advantages.

The mistake would be assuming that because those advantages are valuable, the operating model built around them should remain unchanged.

The opportunity is to combine what an established company already has with what an AI-native company can now do.

That may prove to be far more powerful than either one alone.

I Think Companies Need Three AI Questions

I am becoming less interested in asking only:

Where can we use AI?

It is still worth asking, but I think leadership teams need to look at AI through three different lenses.

Optimize

Where can AI make the existing company faster, less expensive, more accurate, safer, or easier to operate?

This is where many companies are today.

It includes copilots, better analysis, automated reporting, improved search, faster customer service, and less manual work.

There is plenty of value here.

Reinvent

The next question is harder:

If machines can now retrieve information, reason about it, communicate, and take authorized actions, how should this process work?

This is not about automating every step of an old workflow.

Sometimes the right answer will be to remove steps entirely.

A monthly management report might become continuous exception monitoring. A handoff between departments might become a system-to-system action. A salesperson searching for opportunities might become an agent continuously identifying them.

That is reinvention.

Disrupt

Then comes the most uncomfortable question:

What becomes possible now that was not practical before AI?

Maybe a competitor could serve customers with a much smaller organization.

Maybe parts of an industry that exist mainly to coordinate information become less important.

Maybe buying and selling begin to happen through agents.

Maybe the economics of a service change because expertise that used to require many hours of human labor can remain inside the customer’s business as software.

We should not assume all of these things will happen.

But I think established businesses should at least ask the questions.

The 22-Year-Old Question

Here is an exercise I think almost any leadership team could try.

Pick an important part of your company.

It could be sales, purchasing, customer service, pricing, accounting, distribution, scheduling, reporting, or something else.

Then imagine a smart 22-year-old AI entrepreneur decides to compete with you.

She has no existing organization to protect.

She has no legacy systems.

She has no process manual telling her how the industry is supposed to work.

She has no employees telling her, “We have always done it this way.”

She starts with a blank sheet of paper.

Then ask:

If she were building a competitor to my company today, knowing what AI can do, why would she build it the way we did?

Which parts would she keep?

Which parts would she eliminate?

Which decisions would become continuous?

Which customer interactions would change?

Which jobs would become different?

Which parts of your value would remain extremely difficult to reproduce?

And which parts only exist because, until recently, humans had to move the information around?

I suspect the answers could be uncomfortable.

They could also be extremely valuable.

AI May Change More Than Productivity

For the last couple of years, much of the AI discussion has been about what AI can do for us.

That is important.

But the bigger business story may eventually be what AI allows someone to build that could not have existed before.

The Internet gave companies new tools. It also gave us Amazon.

AI will give established companies extraordinary new tools.

But I think we should prepare for the possibility that it will also create companies that look very different from the ones we know today.

The question for established companies is not simply whether they will adopt AI.

It may be whether they can use AI to rethink their own business before somebody else does