Every time a startup with a “weird” mission emerges in Silicon Valley, people have a hard time taking it seriously or thinking it’s satire. This time, it’s no different.
Tamay Besiroglu, a renowned AI researcher and founder of the non-profit research organization Epoch, launched a startup called Mechanize. He announced the startup’s main goal as “the complete automation of work” and “the complete automation of the economy.” In other words, the startup plans to transfer almost every human job to AI agents.
Market potential: $60 trillion?
Besiroglu estimated Mechanize’s target market as not just one industry, but the annual wage paid to all of humanity — $60 trillion. In the United States, that figure is $18 trillion.
He believes that this new era will begin with “white collar workers” (office workers, analysts, writers). Because robotics will be needed to automate physical work, which is more complex.
Negative reaction from the public and academic colleagues
The startup also posed a threat to the reputation of the Epoch Institute. The institute’s director even quipped in X: “The best birthday present I’ve ever gotten: a PR crisis.” Many in the scientific community say that the knowledge gained through Besiroglu’s Epoch research, which was previously considered “neutral and reliable,” is now being questioned.
Who is backing it?
Mechanize has several prominent investors — Nat Friedman, Daniel Gross, Patrick Collison, Jeff Dean, Dwarkesh Patel. While some have yet to comment on the matter, investor Marcus Abramovitch has described the idea as “coming from some of the deepest thinkers in humanity.”
“Is it good for humanity?”
Besiroglu believes this approach will lead to “explosive economic growth.” Some workers will find value in “AI-enabled jobs.” Even if wages decline, they can survive on dividends, rental income, or government assistance.
But critics point to a simple truth behind this idea: who will buy products and services if people can’t buy them themselves?
The technical truth: agents still don’t work well
Besiroglu himself admits that agents in their current form are unreliable, have no memory, can’t work independently, and can get lost in the pursuit of long-term plans.
But he’s ready to solve this problem — hundreds of agent startups, including Microsoft, Salesforce, OpenAI, and others, are working on it.
Tamay Besiroglu is determined to automate the world. Only time will tell whether this approach will bring us time, prosperity, and opportunity, or whether it will become a cause of global unemployment, social inequality, and unrest.
Can AI agents really do the work of humans in full?
