Eric Schmidt rarely talks like a founder chasing the next breakout. Google’s former CEO has a careful public style, so one line at the Imagination in Action AI Summit in Davos stood out for how direct it was. His advice for making money in AI was to found a company built on AI agents, software that carries out tasks by itself. Forbes captured the practical version, which was to build an agent to do something.
Schmidt’s advice breaks with how software companies have long worked. They built a product and sold access to it, leaving customers to turn it into value with their own teams and processes. A company built on agents starts with the work itself and can operate faster, because its agents do that work inside the business.
That idea applies across very different businesses. An energy company could use agents to price supply and demand in real time, and a healthcare company could use them to support drug discovery and diagnosis. A record company could be built around AI artists.
The plainest example is a roofing company that uses agents to quote, qualify, close and invoice customers faster than a traditional operator. The company still sells roofing, and its advantage lies in the work behind the sale. Faster quotes can help win customers, faster qualification can reduce wasted time, and faster invoicing can improve cash collection.
Work like that once required more people. Because intelligence can be programmed, that work can be designed around systems that act, which points founders toward new companies instead of small AI add-ons.
In the same Davos discussion, Schmidt described the moment as the arrival of non-human intelligence that can compete with people. He said the outcome depends on how humans use it, shape it and make sure it stays pointed at human intent. On that view, companies built on that intelligence have to match the size of the moment.
He also said the year or two after Davos would be full of people building agents, and those agents would compete. If one founder builds a better agent than anyone else, he argued, that founder could build one incredible company and own it.
