The modern corporate hierarchy is essentially a game of telephone. An idea starts at the top, gets filtered through three layers of VPs, is diluted by middle managers, and arrives at the engineer's desk as a distorted version of the original intent. For decades, this was the only way to scale a company, as human managers were the only reliable mechanism for distributing information and enforcing alignment. But Jack Dorsey is betting that the era of the middle manager is over, and he is using his company, Block, as the primary laboratory for this hypothesis.
The Architecture of a Flat Intelligence
Block has officially moved beyond the phase of treating AI as a mere productivity booster. In a radical restructuring, the company has reduced its total workforce by 40%, a move that Dorsey frames not as a cost-cutting exercise, but as a fundamental redesign of the company's operating system. The goal is to transition from a traditional hierarchical organization into what Dorsey calls an intelligence. This new structure is designed to eliminate the information bottlenecks that have plagued corporate management for two millennia.
At the center of this transition is a drastic reduction in organizational depth. Block is currently compressing its management layers from five levels down to two or three. The ultimate objective is a flat structure where 6,000 employees report directly to the CEO. To make this feasible, the company is replacing the human relay system with an AI-driven intelligence layer. By integrating every corporate artifact—including emails, Slack messages, code pull requests, and meeting minutes—into a centralized AI model, Block is creating an environment where any employee can converse directly with the company's collective knowledge base.
This shift has necessitated a complete redefinition of professional roles within the firm. The traditional manager has been phased out in favor of three specific designations. First are the ICs, or Individual Contributors, who serve as the primary builders and operators. Second are the DRIs, or Directly Responsible Individuals, who hold direct accountability for customer outcomes. Finally, the company has introduced the Player-Coach. Unlike a traditional boss, the Player-Coach does not exercise command-and-control authority but is instead assigned to provide flexible support and skill development to ICs and DRIs.
The technical catalyst for this upheaval arrived in December 2024. Block determined that coding models had finally reached a critical threshold of maturity, becoming capable of managing and interpreting legacy codebases without constant human hand-holding. This capability allowed the company to employ a backcasting strategy, calculating the absolute minimum number of personnel required to maintain services and ensure regulatory compliance, and then stripping away the layers of management that previously existed to coordinate those people.
From Persuasion Slides to Live Prototypes
This structural collapse changes the very nature of how work is validated at Block. For years, the corporate meeting was a theater of persuasion. Teams would spend hours crafting Google Docs and Google Slides to argue the logical validity of a proposal, using diagrams and bullet points to convince leadership to greenlight a project. In the new intelligence-centric model, the slide deck is dead. Meetings now revolve around live, functioning prototypes driven by real-time data.
This shift is a direct result of the intelligence layer. Because the AI has modeled the company's internal data and customer signals, the cost of exploration has plummeted. When the cost of testing a hypothesis drops toward zero, the need for a middle manager to vet a proposal vanishes. Instead of discussing a theoretical plan, teams now demonstrate a working model and observe the data signals in real-time. This replaces the interpretative layer of management—where a director might summarize a report for a VP—with a direct, unmediated line of sight between the data and the decision-maker.
This transition also redefines the value of the human worker. In a world where AI can handle the bulk of implementation and legacy maintenance, the developer's value is no longer found in the ability to write a specific function. Instead, the premium has shifted to orchestration. The core competency is now the ability to analyze AI output, treat it as a raw data point for a better input, and iteratively refine the prompt to reach a superior result. The workflow has moved from a linear path of input-and-output to a recursive loop of constant reprogramming.
This new reality extends to leadership through what is known as the ALE frame, which combines Authenticity, Logic, and Empathy. As the administrative functions of leadership—tracking tasks, reporting status, and distributing information—are absorbed by the AI layer, the human leader is stripped of their traditional tools of control. What remains is the need for psychological safety and logical direction. The leader's role is no longer to manage the work, but to coach the person. This is why the Player-Coach role is focused on empowerment rather than oversight.
However, this model is not a universal blueprint. It relies heavily on a business model where customer signals are clear and immediate. Block's ability to feed direct customer feedback into an AI model, which then informs the product roadmap without passing through five layers of committee, is what gives the intelligence structure its speed. The company is essentially treating itself as a piece of software that can be continuously refactored.
The success of this experiment will determine if the corporation of the future is a pyramid of people or a network of agents orbiting a single, shared intelligence.
