In Jack Dorsey’s view of the world, the job most at risk from the AI revolution is the middle manager.

Dorsey argues in a new essay, “From Hierarchy to Intelligence,” published with Roelof Botha, Sequoia Capital’s managing partner, an investor in Block, that his company’s decision to cut approximately 4,000 of its more than 10,000 employees was not a cost reduction but a permanent restructuring to replace middle managers with AI.

Corporate hierarchy, the essay argues, has always existed to solve one problem: routing information through organizations too large for any single person to oversee.

Managers aggregate context from below, act as messengers from above, and maintain alignment across teams. AI can now perform those functions continuously and at scale, the authors argue, making the messenger redundant.

In place of management layers, Dorsey and Botha proposes two AI-driven “world models.”

One aggregates internal data from code, decisions, workflows, and performance metrics to create a continuously updated picture of company operations, replacing the context that managers traditionally carried.

The other maps customer and merchant behavior using transaction data from Cash App and Square.

Those models feed what Block calls an “intelligence layer” that composes financial products dynamically to fit market demand.

If done properly, the models absorb the coordination work that previously justified the existence of middle management.

Rather than building from fixed roadmaps, the essay proposes breaking Block’s business into modular capabilities, including payments, lending, card issuance and payroll.

When the system identifies a need, the essay’s example is a merchant facing a seasonal cash flow gap, it assembles a solution from existing capabilities. When it cannot, the missing capability defines what gets built next, replacing the product roadmap with a system-generated backlog.

The organizational structure is reduced accordingly. Block plans to operate with three roles: individual contributors who build the system, directly responsible individuals who own specific outcomes on 90-day cycles, and player-coaches who remain hands-on while developing people.

Dorsey told Wired in early Marchthe restructuring was triggered by a capability shift he observed in December in tools including Anthropic’s Opus 4.6 and OpenAI’s Codex 5.3, which he said was now capable of operating effectively in large codebases.

But current and former Block employees told the Guardian that roughly 95% of AI-generated code changes still require human modification, and that AI tools cannot yet lead in regulated areas like banking and money transfers.