Zuckerberg and Dorsey Want to Clone Themselves With AI to Run Companies From Everywhere

Mark Zuckerberg and Jack Dorsey have independently arrived at the same idea: AI can become an extension of the CEO, acting, making decisions and keeping a finger on the pulse where the executive physically cannot be. The concept rests on AI agents trained on a specific executive's way of thinking, priorities and decisions. They can review code before it goes to review, sit in planning meetings, answer internal requests and analyze team metrics, all on behalf of, or in the spirit of, their "original."

For a large company where a CEO physically cannot cover everything, this looks like an appealing fix for the problem of scale. Zuckerberg has long said Meta is moving toward a model in which AI does part of the work of junior specialists. The next step is an AI that represents the CEO himself: it takes part in product discussions, conveys his views to the team and assesses whether initiatives fit the strategy.

This is not automation of routine work; it is an attempt to scale an executive's personality. Dorsey, for his part, is known for radical views on how work should be organized. At Block he is experimenting with decentralized teams where decisions are made quickly and without layers of bureaucracy.

An AI agent as the "voice of the founder" fits that logic: there is no need to wait for a meeting with the CEO, because the system already knows what he would say. Despite the differences in style, one thing unites both visions: stronger control. Not in the sense of surveillance, but in the sense of managerial reach: the CEO gets to be "everywhere" without being physically present.

This changes the very nature of corporate hierarchy. If an executive's influence was once limited by time and attention, that barrier is now starting to disappear. Critics point to the downside: when AI speaks "on behalf of" a leader, there is a risk of distortion, of losing live contact with the team, and of an illusion of presence where a real human decision is needed.

Employees may get not access to the CEO but only to his simulacrum, accurate enough to replace him but not alive enough to notice nuance. Still, the trend is clear. The concept of "AI as an extension of the leader" is the next stage after AI assistants and AI employees.

The question is no longer whether CEOs will use AI to manage, but how transparently they will tell their teams about it, and where the line falls between augmentation and replacement.

*Meta has been designated an extremist organization and is banned in Russia.