Will Middle Management Be Replaced by AI? What Will the Future Company Structure Look Like

marsbit发布于2026-04-01更新于2026-04-01

文章摘要

The article explores whether AI will eliminate middle management and reshape future corporate structures. It traces the historical evolution of organizations—from Roman military units to modern corporations—showing how hierarchical systems emerged to manage information flow under the constraint of limited "span of control." Middle management, matrix structures, and bureaucratic systems were all solutions to coordination challenges in information-scarce environments. AI, however, challenges this foundational premise. By enabling real-time modeling, understanding, and distribution of information, AI could replace human-centric coordination mechanisms. Examples like the AI firm "Moon Dark Side" illustrate radical experiments: no departments, titles, or traditional KPIs, with co-founders directly managing large teams and AI agents handling tasks from data processing to code generation. Block (founded by Jack Dorsey) is presented as a case study in building an "intelligent company." This model relies on two core components: a "company world model" (a real-time understanding of internal operations via digital traces) and a "customer world model" (built from real behavioral data, especially financial transactions). An intelligence layer uses these models to dynamically combine capabilities (e.g., payments, lending) to serve customers proactively, without pre-defined product roadmaps. In this structure, traditional roles shift. Middle managers are replaced by a system that handles...

Editor's Note: While most companies still view AI as a "productivity tool," Jack Dorsey has taken the question a step further: Is AI rewriting the very logic of how organizations operate? As the co-founder and former CEO of X and the founder of Block, he has long focused on the relationship between technology and organizational forms.

This article starts from history, re-examining why enterprises have evolved into their current forms and why this structure is beginning to loosen. From the Roman legions to modern corporations, organizational evolution over the past two millennia has always revolved around the same constraint: achieving information transmission and coordination within a limited "span of control." Hierarchical structures, middle management, and matrix systems are essentially different solutions to this problem.

The emergence of AI, for the first time, challenges this premise. When information can be modeled, understood, and distributed in real-time, does an organization still need a human-centric coordination mechanism?

Similar changes are already appearing in reality. Recently, the publication Renwu reported that the AI company "Moon Dark Side" operates a team of over 300 people with no departments, no job titles, and no OKRs or KPIs. Collaboration relies on direct communication rather than layered reporting; five co-founders each directly manage 40–50 employees. Meanwhile, Agents are embedded into daily workflows, capable of completing tasks like information organization, product design, and even code generation in a short time. This structure is not simply about "removing management" but rather shifting the complexity upfront into recruitment, fluidity, and tool systems.

Using Block's practices as a starting point, this article further proposes a more radical vision: moving from a "hierarchical organization" to an "intelligent company," replacing traditional information routing systems with a "company world model + customer world model + intelligence layer," and even redefining middle management itself. This is not just an issue of efficiency but potentially a rewrite of organizational forms.

Below is the original text:

In the view of Sequoia Capital, "speed" is the best indicator for predicting the success of a startup. Most companies still see AI as a tool to enhance productivity, while only a few have begun to focus on how AI changes the way people collaborate. Block is demonstrating a全新的 path: fundamentally rethinking organizational design and using AI as a compound competitive advantage that continuously amplifies "speed."

The Origin of Hierarchical Organizations: From Roman Legions to Modern Corporations

Two millennia before the corporate organizational chart appeared, the Roman army had already solved a problem that still plagues large organizations today: how to coordinate thousands of people with limited communication and over vast distances.

Their solution was to establish a nested command system with relatively stable "spans of control" at each level. The smallest unit was the "contubernium," consisting of 8 soldiers who shared a tent, equipment, and a mule, led by a decanus. 10 contubernia formed a "century" (actually about 80 men), commanded by a centurion; 6 centuries constituted a cohort; and 10 cohorts formed a legion of about 5000 men.

At each level, there was a clear commander responsible for aggregating information upward and conveying orders downward. This structure from 8 → 80 → 480 → 5000 was essentially an efficient information transmission mechanism, built on a simple yet crucial premise: the number of people one person can effectively manage directly is typically only 3 to 8. The Romans gradually discovered this rule through prolonged warfare. Even today, the U.S. military's hierarchical system largely follows a similar logic. We call this constraint the "span of control," and it remains a fundamental limitation that all large organizations must contend with.

The next major变革 came from Prussia.

After a crushing defeat by Napoleon at the Battle of Jena in 1806, Scharnhorst and Gneisenau led military reforms, proposing an uncomfortable reality: one cannot rely on individual genius; one must rely on systems. They established the "General Staff," training a class of专职 officers whose duty was not to fight but to plan operations, process information, and coordinate across units. Scharnhorst's original intention was to "compensate for the deficiencies of incompetent generals, providing them with the abilities they lack." This was essentially the雏形 of "middle management": a group of professionals responsible for information transmission, pre-calculating decisions, and maintaining the coordination of complex organizations. Simultaneously, the military clearly distinguished between "line" and "staff" functions: the former推进 core tasks, the latter provided professional support. This division is still widely used in businesses today.

In the 1840s and 50s, American railroad companies introduced the military hierarchical system into the business world.

The U.S. Army supplied railroad companies with大量 engineers trained at West Point, who brought military organizational thinking with them. Line and staff structures, divisional划分, and bureaucratic reporting and control systems all originated in the military. In the mid-1850s, Daniel McCallum of the New York and Erie Railroad drew the world's first organizational chart to manage a 500-mile railroad system and thousands of employees. The previous informal management methods suitable for small railroads had failed, leading to frequent train collisions. McCallum institutionalized the Roman-style hierarchical logic: clear levels of authority and responsibility, defined reporting relationships, and structured information flow. This became the prototype for the modern corporation.

Subsequently, Frederick Taylor (known as the "father of scientific management") optimized the internals of this system. He broke down work into specialized tasks, assigned them to trained experts, and managed with quantitative metrics rather than intuition, thus forming the "functional pyramid" structure—an organizational form that maximized efficiency within the existing information routing system.

The first major stress test for this functional structure occurred during WWII with the "Manhattan Project." The project required physicists, chemists, engineers, metallurgists, and military personnel to collaborate across disciplines under extreme secrecy and time pressure to achieve a single goal. Robert Oppenheimer at Los Alamos Laboratory used functional divisions but insisted on open cross-departmental collaboration, resisting the military's "compartmentalization" tendencies. In 1944, when the "implosion problem" became a critical bottleneck, he reorganized teams, creating cross-functional groups—something almost unheard of in the business world at the time. This model worked, but it was a wartime exception, driven by a few exceptional individuals. The question for the post-war business world was: Could this kind of cross-functional collaboration become常态化?

Post-war corporate growth and global expansion made the limitations of the functional structure increasingly apparent.

In 1959, McKinsey's Gilbert Clee and Alfred di Scipio published "Creating a World Enterprise" in the Harvard Business Review, proposing the "matrix organization" framework, combining functional expertise with divisional structure. With Marvin Bower's push, McKinsey helped companies like Shell and General Electric implement this model, achieving a balance between "central standards" and "local flexibility." This system became the paradigm of the "modern enterprise" in the post-war global economy.

Subsequently, to address the complexity and bureaucratization of the matrix structure, new management frameworks continuously emerged.

McKinsey proposed the "7-S Model" in the 1970s, distinguishing between "hard elements" (strategy, structure, systems) and "soft elements" (shared values, skills, staff, style), emphasizing that structure alone cannot guarantee organizational effectiveness and requires coordination at the cultural and human levels.

In recent decades, tech companies have conducted more radical experiments with organizational structure.

Spotify introduced cross-functional squads and short-cycle iterations; Zappos experimented with Holacracy, eliminating management titles; Valve adopted a flat structure with no formal hierarchy. These attempts all revealed the limitations of traditional hierarchies but failed to completely solve the problem: Spotify returned to traditional management as it scaled, Zappos experienced significant employee turnover, and Valve's model was difficult to scale beyond a few hundred people. When organizations reach thousands of people, they still have to revert to hierarchical coordination because there is no more effective information routing mechanism.

This constraint is exactly the same problem faced by the Romans and the Marines in WWII: a smaller span of control means增加层级, and增加层级 slows down information flow. For two thousand years, organizational innovation has always tried to circumvent this trade-off but has never truly broken it.

So, What's Different Now?

At Block, we have begun to question a fundamental assumption: that organizations must use humans as the coordination mechanism, adopting hierarchical structures. Our goal is to replace the functions of hierarchy with systems. Currently, most companies are just equipping employees with AI co-pilots, making existing structures run a bit better, but the essence remains unchanged. What we want to build is another form: a company that is itself an "intelligent agent" (or even a small AGI).

We are not the first organization to try to move beyond hierarchy. Haier's "Rendanheyi," platform organizations, "data-driven management,"等都是 similar explorations. But they lack a key element: technology that can truly承担 coordination functions. AI is that technology. For the first time, a system has emerged that can continuously maintain a model of the entire enterprise's operation and coordinate based on it, without humans transmitting information through hierarchies.

To achieve this, a company needs two things: a "world model" of its own operations, and sufficiently rich customer signals.

Block operates remotely, and all work leaves a recordable "trace": decisions, discussions, code, designs, plans, problems, and progress. These constitute the raw materials for the company's world model.

In traditional companies, managers are responsible for understanding team status and transmitting information up and down;而在一个"machine-readable" organization, AI can continuously build this global view: what is being done, where things are stuck, how resources are allocated, what works, what doesn't. This information, previously carried by hierarchy, is now carried by the model.

But system capability depends on the quality of input signals, and "money" is the most真实 signal. People might lie on surveys, ignore ads, abandon shopping carts, but when they spend, save, transfer, borrow, or repay, these actions are real. Block sees both sides of a transaction simultaneously daily: the buyer through Cash App and the seller through Square, along with merchant operational data. This allows it to build a rare customer world model—an understanding of financial behavior per customer and merchant based on real transaction signals, and these signals continuously accumulate and strengthen.

The company world model and the customer world model together form the foundation of a new type of company. In this model, the company no longer operates with product teams围绕既定 roadmaps, but is built around four cores:

First, capabilities: foundational financial capabilities like payments, lending, card issuing, banking, buy-now-pay-later, payroll, etc. These are not products but underlying modules, with no interface, but possessing reliability, compliance, and performance requirements.

Second, world model:包括 the company model (understanding its own operations) and the customer model (built on transaction data, representing customers and markets), gradually evolving into a system with causal and predictive capabilities.

Third, intelligence layer: at specific moments, for specific customers,组合 capabilities to actively provide solutions. For example, when the system predicts a restaurant's cash flow is about to tighten, it automatically组合 loan and repayment options and pushes them in advance; or when user behavior changes suggest they are moving, it automatically configures new financial service combinations. None of this requires prior design by product managers.

Fourth, interfaces: like Square, Cash App, Afterpay, TIDAL, etc. These are just delivery interfaces; real value is generated by the model and intelligence layer.

When the system attempts to组合 a solution but finds it lacks a certain capability, this "failure signal" becomes the future product roadmap. The traditional method of product managers envisioning needs is directly replaced by real customer behavior.

In this structure, the organization也随之 changes. In traditional companies, intelligence is distributed among people and routed by hierarchy; here, intelligence resides in the system, and people are at the "edge." The edge is where intelligence meets reality. People can perceive intuition, culture, trust, and complex situations that the model cannot capture, and they play a role in ethical and high-risk decisions. But they don't need to coordinate through hierarchy because the world model provides the necessary context.

In practice, the organization will simplify into three types of roles:

· IC (Individual Contributor): experts who build capabilities, models, and interfaces;

· DRI (Directly Responsible Individual): mobilizes resources around specific problems or customer outcomes;

· Player-coach: participates in frontline work while also cultivating talent, replacing traditional managers.

Fixed middle management layers are no longer needed; the remaining coordination work is done by the system.

Block is currently still in the early stages of this transition, which will be a difficult process, and some attempts may fail. But we are公开 this direction because we believe every company will eventually face the same question: Are you continuously deepening your understanding of a complex problem?

If the answer is no, AI is just a cost-cutting tool; if the answer is yes, AI will reveal the true essence of the company.

Block's answer is the "Economic Graph": connecting millions of merchants and consumers, understanding the behavior on both sides of transactions in real-time, and continuously accumulating. We believe this model of "organizing a company with intelligence rather than hierarchy" will reshape how various enterprises operate in the coming years.

The speed of a company fundamentally depends on the speed of information flow. Hierarchy and middle management slow down this flow. For two thousand years, from the Roman army to modern enterprises, we had no better alternative. But now, this premise is changing. Block is building the next form.

相关问答

QWhat is the core constraint that has shaped organizational structures for the past two thousand years, according to the article?

AThe core constraint is the limited 'span of control'—the number of people one person can effectively manage and coordinate, which has historically been between 3 to 8 people. This limitation necessitated hierarchical structures for information routing and coordination.

QHow does the article suggest AI is fundamentally change the traditional role of management layers in a company?

AAI is proposed to replace the traditional information routing function of management layers. By building a 'company world model' and a 'customer world model,' an AI-driven 'intelligence layer' can coordinate tasks, allocate resources, and make decisions in real-time, reducing or eliminating the need for human middle managers to facilitate communication and coordination.

QWhat are the four core components that the article states will form the foundation of a new type of 'intelligent company'?

AThe four core components are: 1. Capabilities (underlying functional modules), 2. The World Model (comprising a company model and a customer model), 3. The Intelligence Layer (which combines capabilities for specific customer needs), and 4. Interfaces (the delivery surfaces for the solutions created by the intelligence layer).

QWhat historical organizational structure from the military is cited as a direct precursor to the modern corporate hierarchy?

AThe hierarchical structure of the Roman army, with its nested command units (Contubernium → Century → Cohort → Legion) and a stable span of control, is cited as the precursor. This structure was later formalized in business by figures like Daniel McCallum of the New York and Erie Railroad.

QWhat new organizational roles does the article predict will emerge in an AI-driven 'intelligent company,' replacing traditional middle management?

AThe article predicts three primary roles will replace traditional middle management: Individual Contributors (ICs) who are domain experts, Directly Responsible Individuals (DRIs) who mobilize resources around specific problems, and Player-Coaches who both perform hands-on work and mentor others.

你可能也喜欢

红杉对话黄仁勋:计算模式迎来60年巨变,你不会被AI取代,但会被“善用AI的人”降维打击

来源:红杉资本 编译:Yuliya, PANews 红杉资本合伙人康斯坦丁·布勒与英伟达创始人兼CEO黄仁勋展开对话,探讨计算技术的根本变革。黄仁勋指出,计算模式正经历60年来的巨变:从过去的数据存储与检索,转向实时生成与定制。未来的每个像素、声音、视频都将根据用户上下文实时生成,这催生了“AI工厂”——大规模生产智能的计算设施。 黄仁勋将AI工厂比作新时代的“发电机”,它输入电能,输出数字化的智能(Token)。他预测,未来智能网络将像电网和互联网一样包裹全球,不仅人类使用,还会有上千亿个AI智能体在互联网上协作工作。 针对AI对就业的影响,黄仁勋反驳了“AI取代人类”的恐慌论。他强调,自动化提升的是效率,而非消灭职业。他以放射科医生和软件工程师为例:AI辅助反而扩大了行业规模与人才需求。真正的风险在于,“人们不会因AI失去工作,但可能被善用AI的人取代”。他鼓励所有人积极拥抱AI,将其作为能力升维的工具。 在投资层面,黄仁勋提出“五层蛋糕”模型:能源、芯片与计算设施、基础设施、模型层、应用层。AI生态当前年投入约1万亿美元,未来规模可能达20万亿美元,涵盖能源、硬件、软件及各行业应用,创造大量新机会。 黄仁勋总结,AI是赋予人类超级能力的技术,当前关于失业的叙事多为误导。企业与国家应聚焦AI带来的确定性机遇,加速融入智能时代。

marsbit47分钟前

红杉对话黄仁勋:计算模式迎来60年巨变,你不会被AI取代,但会被“善用AI的人”降维打击

marsbit47分钟前

“我不需要更好的模型了”:Reddit 热帖下的 AI 众生相

Anthropic发布了新一代旗舰AI模型Claude Fable 5,在基准测试中表现卓越,但用户反响冷淡。Reddit上一篇题为“我不需要更好的模型了”的热帖引发广泛讨论,反映了当前用户的主流情绪。 许多用户认为,现有模型(如Opus 4.8)的能力已足够应对日常工作和代码编写,对新模型带来的边际效益提升感觉不明显,同时对其更高的使用成本和Token消耗感到不满,认为AI模型能力可能已进入平台期。 此外,Fable 5内置的偏保守的安全护栏成为主要槽点。用户抱怨在处理与安全相关(如代码安全审查)的请求时,触发拒绝的频率远高于官方宣称的5%,导致请求被降级回旧模型处理,付费用户对此尤其不满。 然而,也存在反对声音,主要来自处理极端复杂任务的深度用户。他们表示,在需要超长上下文理解、处理万行级别代码或复杂仿真的场景中,Fable 5的能力提升是“革命性”的,带来了实质性的效率突破。 这场争论揭示了AI行业的一个现状:基准测试衡量的是能力上限,而大众市场更关注日常需求的天花板。当多数普通用户的需求已被现有模型满足,更强模型的价值主要在特定专业领域凸显。文章最后指出,未来公开模型的发展,将取决于厂商如何平衡能力、成本、安全与可用性,以及重度用户是否愿意为之买单。

marsbit54分钟前

“我不需要更好的模型了”:Reddit 热帖下的 AI 众生相

marsbit54分钟前

当 AI 流量超越人类,如何证明你是真人?

随着AI智能体产生的网络流量超越人类,互联网商业模式面临严峻挑战。AI行为模式与人类截然不同,它们高效抓取数据却不产生广告浏览或消费行为,导致网站无法获利。同时,AI搜索摘要大幅分流原始网站流量,内容创作者陷入被爬虫抓取内容又被AI截留用户的双重困境。 传统验证码等人机区分手段因AI图像识别能力超越人类而失效。行业转向行为生物识别技术,通过监测光标移动、打字节奏、按键力度等人类独有且难以模仿的生理行为特征来验证真人。这类技术甚至能通过行为异常识别诈骗。然而,这引发了谁将掌控用户生物特征数据的隐私担忧。 目前真人验证体系主要有两大方向:一是以Worldcoin为代表的中心化方案,通过扫描虹膜等生物特征生成身份凭证,但存在数据过度采集和管控风险;二是基于加密技术的零知识证明方案,旨在不泄露个人信息的前提下证明“真人”身份,但可能因经济激励而被身份租赁黑市钻空子。 尽管两者均有缺陷,但文章作者倾向于支持加密方案,因为中心化方案可能导致用户的生物数据被永久锁定在采集公司手中,形成无法删除的监控体系。在AI流量日益主导的背景下,如何在有效区分人机与保护用户隐私之间取得平衡,已成为亟待解决的难题。

marsbit1小时前

当 AI 流量超越人类,如何证明你是真人?

marsbit1小时前

去中心化 AI 2026 全景地图:为什么区块链是 AI 绕不开的「解药」?

文章探讨了去中心化AI在2026年的发展全景,并论证了区块链技术对解决中心化AI结构性瓶颈的关键作用。中心化AI面临四大瓶颈:计算资源稀缺昂贵、控制权过度集中、模型输出不可验证,以及训练数据获取困难。区块链技术通过提供开放、可验证且经济可及的解决方案,成为AI发展的必要“解药”。 文章将去中心化AI技术栈分为三层: 1. **应用与服务层**:聚焦于“代理金融”和“代理支付”,允许AI代理将自然语言指令转化为链上交易和自主支付,例如@gizatechxyz、@Infinit_Labs等项目已在处理数十亿美元的交易量。 2. **中间件层**:解决AI代理间的协调、身份和信任问题,例如@GoKiteAI构建的专用L1、@virtuals_io的代理经济操作系统,以及明星项目Bittensor通过子网网络和代币经济激励有用工作。 3. **基础设施层**:包括去中心化计算、推理、训练、数据存储和隐私验证。项目如@akashnet、@ionet提供廉价计算资源;@OpenGradient、@PhalaNetwork等确保推理的可验证与隐私性;@Filecoin、@grass等提供低成本数据存储与采集。 趋势显示,AI需求增长快于基础设施,代理经济成为增长引擎,计算正变为资产类别,代币经济学在协调资源中显现结构性优势。尽管领域仍处早期且采用不均,但Bittensor、NEAR、Base等项目的成功表明,去中心化AI正从叙事演变为协调计算、数据和智能的新模式。

marsbit1小时前

去中心化 AI 2026 全景地图:为什么区块链是 AI 绕不开的「解药」?

marsbit1小时前

a16z Crypto 合伙人:现金流就是护城河

a16z Crypto合伙人Jason Rosenthal认为,现金流本身就是最强大的护城河。历史上最成功的企业往往将自己置于“资金流”的关键节点,通过促进价值在网络中的创造与转移并抽取部分价值而成长。加密货币是首个为此原生构建的现代技术,它使资金能以互联网速度全球流动,并具备可编程性。 区块链本质上是网络型业务,天然具备网络效应。设计良好的代币能协调用户、开发者等各方利益,推动网络发展,并将价值按贡献分配回参与者。这种模式并非新生事物,从历史上的铁路、石油、电信,到如今的谷歌、Meta、AWS,成功的关键都是占据价值流动的枢纽位置。 传统金融服务(如支付、托管、借贷)利润丰厚,但其中存在大量可压缩的低效成本,这为加密初创公司提供了颠覆机会。加密创始人可以构建更高效、可编程、即时全球结算的新版本。此外,机会远不止于金融,在计算、AI数据、能源、太空等新兴领域,由于没有传统基础设施的包袱,更是构建全新资金流业务的蓝海。 文章建议创始人思考:你的业务是否处于资金流中?收入是否能随网络价值增长而同步增长?目标市场中哪里存在不合理的利润抽取?抓住这些机会,融入新的价值流动浪潮,企业就能随着网络效应累积而持续成长。

marsbit1小时前

a16z Crypto 合伙人:现金流就是护城河

marsbit1小时前

交易

现货
合约

热门文章

加密市场宏观研报:原油飓风、AI巨浪与比特币的十字路口

全球金融市场正经历一场由地缘冲突引发的系统性重估:霍尔木兹海峡封锁导致原油一度暴涨30%,G7紧急释放储备后涨幅收窄,滞胀风险取代通胀成为核心担忧,美元成为“唯一避风港”并逼近100大关,亚太及美股遭遇“黑色星期一”全线重挫;AI领域则冰火两重天,国家发改委提出“十五五”末10万亿规模目标,OpenClaw项目火爆推动概念股狂飙;比特币在宏观风暴中跌破70000美元关键防线。

560人学过发布于 2026.03.12更新于 2026.03.12

加密市场宏观研报:原油飓风、AI巨浪与比特币的十字路口

相关讨论

欢迎来到HTX社区。在这里,您可以了解最新的平台发展动态并获得专业的市场意见。以下是用户对AI(AI)币价的意见。

活动图片