I watched a clip of economist Richard Denniss speaking about AI recently, and something he said has stayed with me.

His concern was about the business model behind it. Companies paying for AI services instead of paying people to do the work.

The more I thought about that, the more uncomfortable I became.

I use AI. I build with it. I’m excited about what it can help people do. But I’m also skeptical of the idea that whatever makes an AI company more valuable must somehow be good for the rest of us.

Those are very different things.

The business model is what worries me

Denniss’s parliamentary remarks, reported by The Point, describe employers buying “AI services instead of labour services.” His broader warning is that expectations of replacing human work are helping drive the enormous valuations of AI companies. That is his argument, not proof of how every investor values every company. But I think it deserves a serious answer. [1]

And when you look at what some companies are selling, it becomes harder to dismiss.

Salesforce openly describes its offering as “digital labor.” Its website quotes CEO Marc Benioff saying, “We’re now providing digital workers.” Intercom charges for outcomes delivered by its Fin AI agent, including customer-service work completed without human involvement. [2][3]

There is a clear commercial proposition here: pay us to perform work.

That can be useful. It can also mean that money which once paid someone’s wages becomes a recurring payment to a technology supplier.

One company’s savings. Someone else’s livelihood.

Imagine a company that can cut its staffing costs substantially by paying for AI and keeping a smaller team. If the numbers work and the service is good enough, I understand why management would consider it. I run a business. Costs matter.

What worries me is what happens when that same calculation is made across thousands of businesses.

The savings belong to the employer. The AI supplier gains a customer. The person whose income disappears has to figure out what comes next.

And it may happen quietly. Someone leaves and isn’t replaced. A contractor stops getting work. An entry-level position never gets advertised. A team takes on more work without hiring anyone.

We should be looking at those changes as well as layoffs.

There is evidence that businesses are using these tools to hand off tasks. In its September 2025 Economic Index, Anthropic classified 77% of sampled business API transcripts as showing automation patterns. That does not mean 77% of jobs disappeared. It means the AI was commonly being asked to perform tasks rather than work through them collaboratively with a person. [4]

I don’t want to exaggerate that finding. I also don’t think we should ignore what it suggests about how businesses are putting AI to use.

There’s money coming in. That isn’t the same as profit.

The financial side adds to my concern.

It would be wrong to say the leading AI companies aren’t bringing in money. OpenAI’s CFO reported more than $20 billion in annualized recurring revenue for 2025. That is a revenue run rate, not profit. The same company describes agents and workflow automation as a next phase of its business, alongside subscriptions, advertising, commerce and other revenue opportunities. [5]

So replacing workers is not the only way these companies can make money.

But it is a very large potential market. If a supplier can take even a small share of what businesses currently spend on human work, that could become a substantial recurring business.

I can see why investors would find that attractive.

I’m less convinced we have thought through what it could mean for everyone else.

If people lose their income, who keeps buying?

People’s wages support other businesses. They pay for groceries, rent, repairs, restaurant meals and children’s activities. In a community like ours, that spending helps keep other people employed.

If enough people lose income, who buys what the increasingly automated companies produce?

I know there are possible answers. Lower prices could help. New businesses and new jobs could emerge. Governments could change how they tax income and support people through the transition.

But “new jobs will appear” feels like an incomplete answer to someone whose livelihood is on the line. What jobs? Where? At what pay? And what happens while they’re waiting?

I don’t know how far this will go. Nobody can honestly promise that mass unemployment is inevitable, or that everything will work out fine.

My worry is that the incentive to replace paid work is already there, while the plan for the people affected is much less clear.

Where Ravenus stands

That is a future Ravenus is against.

For us, useful technology should help an overwhelmed business owner stay organized, help staff spend less time on repetitive administration, or help a nonprofit put more of its energy into the people it serves.

If a tool saves someone several hours, I want that person to benefit too. More time to do good work. Less pressure. A chance to learn. More room to help a customer properly.

I don’t want the first question to be how quickly we can eliminate their position.

I also recognize the tension: Ravenus uses tools from some of the companies whose incentives I’m questioning. Using those tools doesn’t mean accepting every direction their suppliers want to take. It does mean we have a responsibility to think carefully about what we build and how we use it.

We cannot control what every employer does. We can be clear about our own purpose, involve the people affected, and refuse to make replacing them the goal of our work.

The question I keep coming back to

I’m still excited about AI. I’m also worried about where the money is pushing it.

Before we celebrate another business replacing payroll with an AI subscription, I think we need to ask what happened to the people behind those wages.

If they’re worse off, calling it progress isn’t enough.

Sources

1. The Point: Report on Richard Denniss’s parliamentary testimony, October 22, 2026

2. Salesforce: What Is Digital Labor?

3. Intercom: Pricing and outcome-based AI services

4. Anthropic: Economic Index, September 2025

5. OpenAI: A business that scales with the value of intelligence, January 18, 2026

Illustrations are AI-generated conceptual images of fictional scenes.