Author: Jack Dorsey and Roelof Botha, Block
Translation: Yangz, Techub News
At Sequoia, we found that speed is the best predictor of success in entrepreneurship. Most companies view AI as a productivity enhancement tool. However, few focus on AI's potential to change the way we collaborate. What Block is demonstrating is a fundamental restructuring of organizational design, ultimately using AI to enhance speed, making it a compounding competitive advantage.
Two thousand years ago, when the first organizational chart was born, the Roman army solved a problem that still troubles large organizations today: how to coordinate thousands of people dispersed over vast areas with limited communication?
Their answer was a nested hierarchical structure that maintained a stable span of management at each level. The smallest unit was called the "contubernium," consisting of eight soldiers who shared a tent, equipment, and a mule, led by a decanus (ten-man leader). Ten contubernia made up a "century," totaling eighty people, commanded by a centurion. Six centuries formed a "cohort." Ten cohorts made a legion (legion), about five thousand people. At each level, there was a designated commander with clear authority, responsible for aggregating information from subordinates and communicating decisions down the chain. This structure (8 → 80 → 480 → 5,000) is essentially an information routing protocol grounded in a simple human limitation: the number of people effectively managed by one leader is approximately between three and eight. The Romans discovered this over hundreds of years of warfare. Even today, the hierarchical chain of the U.S. Army follows a similar pattern. We now call it "span of control," and it remains the primary constraint for every large organization on Earth.
The next major transformation came from Prussia. In 1806, Napoleon's army destroyed the Prussian army in the Battle of Jena. Subsequently, a group of reformers led by Scharnhorst and Gneisenau rebuilt the army around a disturbing realization: you cannot rely on the genius of top individuals; what you need is a system. They created the General Staff, a specialized layer composed of trained officers whose job was not to fight but to plan operations, handle information, and coordinate various troops. Scharnhorst hoped these staff officers could "assist incompetent generals by providing the talents that leaders and commanders might lack." This was the prototype of what we now refer to as "middle management." The duty of these professionals was to route information in complex organizations, pre-calculate decisions, and maintain alignment. The military also formally differentiated between "line" and "staff" authority. Line departments advance core missions, while staff departments provide professional support. To this day, every company still uses this set of terms.
This military hierarchical structure entered the business world through American railroads in the 1840s and 1850s. The U.S. Army loaned West Point-trained engineers to private railroad companies, bringing military organizational thinking with them. Functions and line hierarchies, divisional structures, and bureaucratic reporting and control systems: all of these were developed in the military and later adopted by railroad companies. In the mid-1850s, Daniel McCallum of the New York and Erie Railroad created the world's first organizational chart to manage a system stretching over 500 miles with thousands of employees. The informal management methods that worked on small railroads began to fail, leading to frequent train collisions and casualties. McCallum's organizational chart formalized the hierarchical logic used by the Romans: power hierarchy, clear reporting relationships, and structured information flow. It became the blueprint for modern companies.
Later, Frederick Taylor, known as the "father of scientific management," optimized the work within this hierarchy. Taylor broke work down into specialized tasks assigned to trained specialists and managed through measurement rather than intuition. This gave birth to the functional pyramid organization, a structure optimized for the efficiency of information transmission, originally pioneered by the military and later commercialized by railroads.
During World War II, the functional hierarchy faced its first real test. The Manhattan Project required physicists, chemists, engineers, metallurgists, and military personnel to collaborate across disciplines under extreme secrecy and time pressure to achieve a common goal. Oppenheimer organized the Los Alamos laboratory into several functional divisions but insisted on open collaboration across divisions, resisting the military's usual departmental division model. In 1944, when the nuclear explosion issue became a critical bottleneck, he reorganized the laboratory around this issue and created cross-functional teams, which were unprecedented in the American business world at that time. This approach worked, but it was an exception led by an extraordinary leader in wartime. The post-war business world faced the question: could this cross-functional collaboration become a norm?
With the growth and globalization of enterprises after World War II, the scaling limitation of functional design became increasingly evident. In 1959, McKinsey's Gilbert Clee and Alfred di Scipio published an article in the Harvard Business Review titled "Creating World-Class Companies," providing a theoretical framework for matrix organizational structures that combined functional expertise with divisional interests. Under the leadership of Marvin Bower, McKinsey helped companies like Shell and General Electric implement these principles to balance central standards with local flexibility. This became the prototype of the "specialized" or "modern" company, driving post-war global economic development.
Over time, other frameworks emerged to address the complexity, rigidity, and bureaucratic issues of matrix structures. In the late 1970s, the McKinsey 7S framework proposed by Tom Peters and Robert Waterman distinguished between "hard S" (strategy, structure, systems) and "soft S" (shared values, skills, staff, style). Its core idea is that relying solely on structural elements is insufficient. Organizational effectiveness requires alignment between cultural traits and the factors that determine whether a strategy can truly succeed.
In recent decades, technology companies have conducted bold experiments in organizational structure. Spotify promoted a cross-functional team model based on short-cycle sprints. Zappos experimented with the holacracy model, entirely eliminating management titles. Valve adopted a flat structure without formal hierarchies. These experiments revealed certain limitations of traditional hierarchical structures, but none solved the fundamental issue. Spotify reverted to traditional management models after scaling. Zappos experienced significant employee turnover. Valve's model proved difficult to scale beyond a few hundred people. When organizations grow to thousands, they all revert to hierarchical coordination because no alternative information routing mechanism has yet proven strong enough to replace it.
This constraint echoes the challenges once faced by the Romans and rediscovered by the U.S. Marine Corps during World War II: narrowing the management span means increasing command levels, but the more levels there are, the slower the information flow. Two thousand years of organizational innovation has essentially aimed to evade this trade-off but has never managed to break it.
So, what is different now?
At Block, we are questioning a fundamental assumption that organizations must be structured hierarchically and rely on humans as a coordination mechanism. Instead, we intend to replace the functions assumed by hierarchies. Most companies using AI today are equipping everyone with a "virtual assistant," slightly improving operations without changing their essence. However, what we pursue is fundamentally different: a company built as an "agent" (or mini AGI).
We are not the first to attempt to transcend traditional hierarchies. Haier's human-single model, platform organizations, and "data-driven" management are real explorations addressing the same issue. What they lack is a technology capable of genuinely executing the coordination functions carried out by hierarchies. AI is that technology. For the first time in history, a system will be able to maintain a continuously updated business model and utilize this model to coordinate work. Previously, this job required humans to relay information through multiple layers of management.
To achieve this, a company needs two things: one is some form of "world model" about its own operations, and the other is sufficient customer signals to make this model useful.
Block prioritizes remote work. Everything we do generates artifacts. Decisions, discussions, code, designs, plans, problems, and progress exist in a documented form. These are the raw materials for building the company's world model. In traditional companies, a manager's job is to understand what is happening in their team and relay that contextual information up and down the hierarchy. In a remote-first company with machine-readable work, AI can continuously construct and maintain this picture: what is being built, what is stuck, how resources are allocated, what is effective, and what is ineffective. This information was previously carried by the hierarchical structure; now it is carried by the company's world model.
However, the system's capability depends on the quality of the customer signals fed into it. And money is the most honest signal in the world.
People might lie in surveys, overlook advertisements. But when they spend, save, transfer, borrow, or pay back money, that is a fact. Every transaction is a fact about someone's life. Block gathers millions of transaction data from buyers and sellers daily through Cash App and Square, along with data generated from merchant operations. This endows the customer world model with a rare ability: an understanding of the financial realities for every customer and merchant built on continuously accumulating honest signals. The richer the signals, the better the model; the better the model, the more transactions; the more transactions, the richer the signals.
The company world model and the customer world model together form the foundation of a new type of company. In such a company, it is no longer the product team building a pre-set roadmap but rather to build four things:
- First: Functions, or atomic financial primitives. Basic financial functions such as payments, lending, card issuing, banking, buy now, pay later, and payroll services are not products; they are hard-to-obtain and maintain foundational components (some with network effects and require regulatory approval). They do not have user interfaces. They have goals regarding reliability, compliance, and performance.
- Second: World Model. This includes two aspects. The company world model enables the company to understand itself and its operations, performance, and priorities, replacing the information flow that previously moved through management hierarchies. The customer world model is based on proprietary transaction data, providing representations of every customer, every merchant, and every market. It begins with raw transaction data and will gradually evolve over time into a complete causal model and predictive model.
- Third: Intelligent Layer. It is responsible for combining basic financial functions into solutions for specific customers at specific moments and proactively delivering them. For example, if a restaurant's cash flow tightens before a seasonal slump, and the model has seen this pattern before, the intelligent layer can combine a short-term loan from the lending function, adjust the repayment plan using the payment function, and present the solution to them before the merchant even thinks to seek financing. Similarly, if a Cash App user's spending pattern changes, the model might associate it with "moving to a new city." The intelligent layer would then combine a new direct deposit setup, a Cash App card optimized for cash back categories in the new community, and a savings goal calibrated according to their updated income. No product manager decided to build these two solutions. The basic financial functions already existed, and the intelligent layer recognized these moments and combined them.
- Fourth: Interfaces (hardware and software). Square, Cash App, Afterpay, TIDAL, bitkey, proto. These are delivery interfaces through which the intelligent layer delivers the combined solutions. They are important, but the core of value creation does not lie here. The value lies in the model and the intelligent layer itself.
When the intelligent layer attempts to build solutions but fails due to missing functions, this failure signal becomes the future product roadmap. The traditional roadmap, which involves product managers hypothesizing the next development steps, is ultimately the limiting factor for any company. In this model, the actual needs of customers directly determine the product backlog.
If this is the product the company builds, then the question becomes: what should employees do?
This gives rise to a new type of organizational structure, fundamentally different from traditional models. In traditional companies, humans are the agents, and hierarchies are responsible for routing various information. In the new model, intelligence resides within the system, with humans positioned on the edge, where action occurs.
The edge is where intelligence intersects with reality. People can delve into areas where the model has yet to reach, perceiving things that the model cannot sense, such as intuition, subjective judgment, cultural background, trust conditions, the atmosphere in the room, etc. They can make decisions that the model should not make alone, especially ethical choices, new situations, and high-risk moments where mistakes can be detrimental. A world model that cannot touch the world is merely a database. But the edge does not require hierarchy for coordination. The world model provides everyone at the edge with the contextual information they need, allowing them to take action without waiting for information to pass up and down the command chain.
In practice, this means we define roles in three categories:
- Individual Contributors (ICs): They build and operate various financial functions, models, intelligent layers, and interfaces. They are deep experts at specific levels of the system. The world model provides the contextual information previously supplied by managers, allowing individual contributors to make decisions at their level without waiting for directives.
- Directly Responsible Individuals (DRIs): They are responsible for specific cross-domain issues, opportunities, or customer outcomes. A DRI might be responsible for a specific merchant churn issue in a 90-day project and has enough authority to mobilize resources as needed from the world model team, lending function team, and interface team. DRIs can continuously focus on certain issues or shift to address new problems.
- People who combine management and hands-on roles: They are involved in the actual building work while also responsible for personnel development. They replace managers who traditionally focused on information routing as their primary responsibility. These individuals still write code, build models, or design interfaces. At the same time, they invest energy into fostering the growth of those around them. They do not spend time in status sync meetings, alignment meetings, or priority negotiations. The world model is responsible for alignment, the DRI structure is responsible for strategy and priorities, and they manage technology and personnel.
There is no longer a need for a permanent middle management tier. Everything else that the old hierarchy did is coordinated by the system. Everyone is empowered, and their roles are closer to actual work and customers.
Block is in the early stages of this transformation. It will be a challenging road, and certain parts are likely to face issues before they truly work. We are writing this now because we believe every company ultimately needs to confront the same question we have faced: what is it that your company understands, but is genuinely difficult to grasp? Is that understanding deepening every day?
If the answer is "nothing," then AI is just a story of cost optimization. You can lay off staff and increase profit margins over a few quarters, but in the end, you will be swallowed by smarter competitors. If the answer is that understanding is deepening, then AI will not merely augment your company; it will reveal the essence of your company.
Block's answer is an economic graph, consisting of millions of merchants and consumers, the two parties in each transaction, and the real-time observed financial behaviors. This understanding compounds with every second the system operates. We believe that the underlying model—a company organized as an agent rather than a hierarchy—holds tremendous significance, enough to reshape the way various companies operate in the coming years. Block's current development is substantial enough to prove that this idea is not merely theoretical (of course, we welcome all discussions and feedback to test and refine our ideas).
The speed of a company's operations depends on the flow of information, and hierarchies and middle management impede that flow. For two thousand years, from the Roman "contubernium" to today's global enterprises, we have never had a true alternative. Eight soldiers sharing a tent need a decanus; eighty people need a centurion; five thousand need a commander. The issue has never been whether you need hierarchy, but rather: for the functions these hierarchies perform, is human coordination the only option? Clearly, it is not anymore. Block is building the organization for the next era.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
