What does AI do to our business model?

For a long time the question was: how do we use artificial intelligence? Now many management teams ask a different one: what does AI do to our business? The current survey by Bitkom, Germany’s digital industry association, shows how quickly the focus is shifting.

31 %expect AI to change their business model (previous year: 24%)

34 %worry that AI threatens their company’s existence (previous year: 23%)

2 in 3AI users have improved their competitive position

What this change looks like in practice mostly stays abstract. Yet it has long been visible. Five signs from around the world show where it is heading. None of them is a distant future. Not every one affects every company. But hardly any business model remains untouched.

Five signs from around the world

01Customers pay for the result, not the hour

The software company Intercom does not bill its AI customer service agent Fin per user license, but per result: 0.99 dollars for every case the AI resolves. If the case is handed over to a person, it costs nothing. The offer became so central to the company that it renamed itself after it: Intercom became Fin. In June 2026, Salesforce announced it would acquire Fin for around 3.6 billion dollars.

Something is changing on the customer side too. The audit firm KPMG threatened to switch its own auditor and pushed the fee down by 14 percent. Its argument: AI makes the work cheaper, so the price has to fall as well. Something similar can be seen at American law firms: clients only accept promises of efficiency if they show up in the price.

For SMBs this means: anyone who bills by effort will soon hear this question, whether an engineering firm, a contract manufacturer or a trade business. Anyone who offers a fixed price for a result, on the other hand, keeps the savings from AI, at least until the competition catches up.

02The machine is paid for by its benefit

The agricultural machinery maker John Deere builds field sprayers whose cameras use AI to recognize weeds. Spraying happens only where weeds grow. On average, this saves almost half of the herbicide. In 2025 the system ran on more than five million acres, around two million hectares. Customers pay per area, and only when the technology saves measurably.

So the manufacturer no longer earns only from the machine, but from what it achieves. It is not an isolated case. In a survey by the audit and advisory firm PwC among just over 400 industrial executives, companies expect to earn around 44 percent of their revenue outside classic product manufacturing by 2030, for example with connected offers, services and outcome-based business models.

For machinery and plant engineering this means: your own knowledge of machine and process becomes the product. Remote monitoring, optimization, availability or guaranteed savings can then be sold as well.

03The customer asks the AI first

Anyone searching on Google more and more often gets an answer that an AI summarizes. When it appears, people click on a result in only 8 percent of searches, compared with 15 percent otherwise. Hardly anyone uses the links in the summary itself. This is shown by an analysis from the Pew Research Center in the United States.

What this can mean for a business model is shown by the American learning platform Chegg. Many students now ask an AI instead of Chegg, and fewer visitors arrive via Google. The company itself names “the new realities of AI” as the reason for the slump. In October 2025 it laid off 45 percent of its workforce.

The next step has already begun. Since September 2025, people in the United States can shop directly in ChatGPT, through an open standard that OpenAI developed with the payment service Stripe. So the AI no longer only recommends, it also completes the purchase.

For SMBs this means: the shortlist of suppliers is more and more often drawn up in a conversation with an AI. Whoever does not appear there makes no list. Product data, references and expertise have to be prepared so that an AI can find and understand them.

04Billing per head comes under pressure

In early 2026 the capital market did a calculation that affects many business models. If AI agents take over office work, companies need fewer employees. And companies with fewer employees need fewer software licenses billed per head. After new AI agents for office work were presented, shares in software, financial services and asset management lost around 285 billion dollars in market value on a single day in early February. At the end of March, an index of listed software companies was down by a good quarter, partly in a general downturn.

The payment provider Klarna shows that replacing people has its limits. In 2024 it said its AI was doing the work of 700 service staff. In 2025 Klarna hired people again, because quality had suffered.

For SMBs this means: offers whose price depends on the number of heads come under pressure, for example at staffing agencies, call centers, training providers and IT system houses. Anyone who buys software, on the other hand, gains negotiating power. And some things that used to be bought as standard software can now be built in-house for less.

05Investors buy traditional businesses to rebuild them with AI

A new kind of buyer is active in the United States. Thrive Holdings, in which OpenAI also holds a stake, buys service providers whose work can be set up anew with AI. They now include more than 70 companies, among them over 50 tax and accounting firms and around 20 IT service providers. Inspection and permitting service providers are next.

AI specialists move into the acquired businesses and rebuild the workflows. According to the company, preparing tax returns has become more than 30 percent faster as a result. In August 2026, investors provided over two billion dollars for this model.

For SMBs this has two sides. Competitors emerge that want to offer traditional services at lower cost. And new buyers emerge. For owners looking for a successor, it becomes important how well their company can develop further with AI. That can influence the purchase price.

What the five signs have in common

Behind all five is the same movement: AI lowers the cost of work. This shifts what customers pay for.

Anyone who sells working time has to renegotiate. Anyone who bills per head loses with every position the customer saves. Anyone who does not appear where customers search today is not even asked. The winners can be those who sell results or have knowledge that nobody else has.

Not every sign hits every company equally, and some are still early. Shopping in ChatGPT exists only in the United States so far, and the figures from Thrive come from the company itself. But the direction is clear. It leads to what we called the core in the lead article of this series: the knowledge your customers pay for and that no competitor can buy. That is where the greatest effect lies, also for new business models.

Change rarely begins with a bang. Usually it begins with a customer who asks about the price. Or with a supplier who does something differently from everyone else. Better to be that supplier yourself.