Mechanize has entered the AI market with a mission that is deliberately sweeping: make it possible to automate all work. The startup, launched by AI researcher Tamay Besiroglu, is drawing attention not only for what it wants to build, but also for what that goal implies for workers, companies, and the research organization Besiroglu founded.
What Mechanize says it wants to build
Besiroglu announced Mechanize on Thursday in a post on X. He described the company’s goal as “the full automation of all work” and “the full automation of the economy.” In practical terms, that means building toward AI systems that could perform jobs now done by people.
The startup is not presenting itself as a robot company focused on physical labor. Besiroglu clarified to TechCrunch that “our immediate focus is indeed on white-collar work” rather than manual labor roles that would require robotics.
Mechanize’s plan centers on the infrastructure needed to make worker automation possible. According to the source article, the company wants to provide data, evaluations, and digital environments that can help automate any job. That makes the startup less a single-purpose AI tool and more an attempt to support broad AI agent capability.
Besiroglu framed the opportunity in unusually large economic terms. He estimated the market by adding up wages paid to human workers, writing that “workers in the US are paid around $18 trillion per year in aggregate. For the entire world, the number is over three times greater, around $60 trillion per year.”
Why the launch became controversial
The backlash began quickly on X. Some critics objected to the mission itself: replacing human work with AI agents is a direct challenge to the role wages play in ordinary economic life. Others focused on the relationship between Mechanize and Epoch, the non-profit AI research organization Besiroglu founded.
Epoch analyzes AI’s economic impact and creates benchmarks for AI performance. The source article says it had been viewed as an impartial way to evaluate performance claims from frontier model makers and others. That reputation is part of why Mechanize’s launch drew scrutiny.
A director at Epoch posted on X, “Yay just what I wanted for my bday: a comms crisis.” The line captured the sense that the announcement had created a reputational problem around the research institute, not just a debate over a new startup.
This is not Epoch’s first controversy. In December, Epoch revealed that OpenAI supported the creation of one of its AI benchmarks. OpenAI then used that benchmark to unveil its new o3 model. According to the source article, social media users argued that Epoch should have been clearer about that relationship earlier.
After Mechanize was announced, X user Oliver Habryka replied, “Alas, this seems like approximate confirmation that Epoch research was directly feeding into frontier capability work, though I had hope that it wouldn’t literally come from you.”
The central tension: abundance or displacement?
Besiroglu argues that automating labor could benefit people by creating “explosive economic growth.” He points to a paper he published on the topic and told TechCrunch, “Completely automating labor could generate vast abundance, much higher standards of living, and new goods and services that we can’t even imagine today.”
That is the optimistic case for Mechanize: if AI agents can take on more work, the economy could produce more. In a narrower version of that idea, human workers might use personal groups of agents to do more than they could do alone.
The concern is equally direct. If people no longer have jobs, they may lose the income they need to buy what those AI systems produce. The source article raises that issue as a basic challenge to the abundance argument.
Besiroglu’s answer is that human wages could rise because people would become “more valuable in complementary roles that AI cannot perform.” But Mechanize’s stated goal is for agents to do all the work, which makes that reassurance complicated.
When asked about scenarios where wages might fall, he said, “Even in scenarios where wages might decrease, economic well-being isn’t solely determined by wages. People typically receive income from other sources — such as rents, dividends, and government welfare.”
That shifts the debate from technical feasibility to economic distribution. If automation produces more goods and services, the question becomes who receives the income, ownership returns, or public support connected to that output.
Investors and the AI agent race
Besiroglu says Mechanize is backed by Nat Friedman and Daniel Gross, Patrick Collison, Dwarkesh Patel, Jeff Dean, Sholto Douglas, and Marcus Abramovitch. TechCrunch reported that Friedman, Gross, and Dean did not return its request for comment.
Marcus Abramovitch confirmed that he invested. The source article identifies him as a managing partner at crypto hedge fund AltX and a self-described “effective altruist.” He told TechCrunch he invested because “the team is exceptional across many dimensions and have thought deeper on AI than anyone I know.”
Mechanize is entering a crowded race. The technical problem Besiroglu is targeting is real: AI agents remain unreliable. He notes that they do not retain information well, struggle to complete tasks independently as requested, and “can’t execute long-term plans without going off the rails.”
Other major companies are also trying to solve agent reliability and usefulness. Salesforce and Microsoft are building agentic platforms. OpenAI is working in the area too. The source article also notes a wider group of startups, including task specialists in outbound sales and financial analysis, companies working on training data, and others focused on agent pricing economics.
What the Mechanize debate reveals
Mechanize is controversial because it says the quiet part plainly. Many companies are trying to make AI agents more capable, but Besiroglu has described the end goal in totalizing terms: full worker automation.
That clarity makes the startup an important test case for the AI economy debate. Its near-term work may focus on white-collar automation, data, evaluations, and digital environments. Its broader ambition, however, raises harder questions about wages, ownership, welfare, and the role of human labor in an economy shaped by AI agents.
For now, Mechanize is both a company and a provocation. It argues that automation can create abundance. Its critics argue that abundance does not automatically mean broad prosperity. The gap between those two claims is where the debate over worker automation now sits.