Are you an AI company, or a company that has an AI slide in the deck?

The question your investors will ask next year is already circulating in the conversations you are not in.

It goes like this: are you an AI company, or are you a software company with AI features?

The practical version sounds like this: do you have salaries that have migrated to software licenses? In other words, is there work in your company that used to require a person and now requires a tool? If yes, you are building toward something. If no, you bought a ChatGPT subscription and called it a strategy.

The distinction matters more than it might seem right now.

Factorial, the HR software company with 10,000 clients across Europe, stopped describing itself to investors as a software company. The claim is not that they added AI to the product. The claim is that AI changed how the company itself operates. That the headcount did not grow at the rate it would have because the machine grew instead. That certain work that used to require people now does not.

Whether you believe the claim or not, the frame is important. The investors and boards asking this question are not asking it to be clever. They are asking it because it predicts something real about which companies are building durable cost and capability advantages and which ones are buying tools.

Block, Jack Dorsey's fintech company, cut 4,000 people from a team of 10,000. Forty percent. The announcement was direct in a way that most corporate communication avoids. AI is doing work that people were doing. The roles did not need to be filled anymore.

That story circulated as a layoff story. It is better understood as an operational transformation story. The 4,000 people did not disappear because business slowed down. They disappeared because the work they were doing found a different home.

For the company with 200 employees, none of this is about mass layoffs. That is not the relevant frame and it is not what the question is really about.

The relevant frame is the next hire.

You have a process that is straining. Or a function that is growing in volume faster than you want to staff it. Or a role that needs to be backfilled. The default response is to hire. That is what you know, it is what worked before, it is what your management team is asking for.

The question is whether the work that role would do is work that a well-designed system could handle instead.

Most mid-sized companies never ask that question. The path of least resistance is to staff. Systems require design, time, someone who understands the work and can translate it into something a machine can execute. Hiring is faster to initiate. It feels more certain. The complexity is familiar.

The cost of this default compounds quietly over years.

The companies that are building toward a different answer are not firing people. They are making different decisions about where the next five roles go. They are looking at each new capacity need and asking whether it belongs in a headcount request or a system design. Some of the time, the answer is still headcount. But the question is being asked, and the answer is sometimes different.

That is the operational difference between an AI company and a software company with AI features.

The tools are not the decision. What you decide when work needs to get done is.

There is a test that is cleaner than any framing exercise. If you disappeared for a month, would the company continue to grow, handle clients, move things forward? Or would it hold its breath until you got back?

If the answer is hold its breath, the question of whether you are an AI company is premature. The more urgent question is whether the company has any systems at all, AI or otherwise, that function without constant human intervention at the center.

Most mid-sized companies do not. The CEO is still the infrastructure. Every decision of any consequence routes through one person, usually the person who built the thing and knows how it actually works. The company scaled the headcount but not the systems. Growth added complexity without adding structure.

That is the thing AI is most useful for in a company like this. Not replacing people. Replacing the CEO as the hidden infrastructure. Taking the judgment calls that route through one person because nobody ever built a system to handle them, and making them structural.

That is a different conversation than should we use AI. That is a conversation about how the company actually operates.

The investor question is a proxy for that conversation. When someone asks whether you are an AI company or a software company, what they are really asking is: are you building a business that operates, or a business that depends on people who operate it?

The honest answer for most companies right now is the second one. The CEO is the infrastructure. Key people are the infrastructure. If they leave, the knowledge leaves. If they are overloaded, the company slows. If the CEO is unavailable, the decision waits.

That is not sustainable at any scale. It was never really sustainable. What changed is the cost of the alternative.

Building systems used to require engineering teams and long timelines and specialized knowledge most mid-sized companies did not have access to. The barrier to building something real has dropped significantly. The question is no longer whether a company your size can afford to build systems. The question is whether you are going to keep choosing not to.

The investor question is coming. Boards are starting to ask versions of it. Clients are starting to ask versions of it. Competitors are starting to answer versions of it.

The version that matters most is the one you ask yourself, alone, after everyone else has gone home.

Are you building a company that runs because of what you built, or one that runs because of who you hired?