The Company Is Alive

By Ayush Kumar

---

Every metaphor we've ever used for a company has been mechanical. An org chart is a wiring diagram. A process is a pipeline. A workflow is an assembly line. Departments are gears. People are resources. We manage headcount the way a factory manages inventory.

This made sense for a hundred and fifty years. Frederick Taylor's scientific management — time every motion, optimize every step, treat the worker as a component — was the right model for an economy where the company's job was to produce. The machine metaphor worked because companies were machines. Inputs went in, outputs came out, managers turned the knobs.

It doesn't work anymore. And the people building the future know it.

Jack Dorsey and Roelof Botha published "From Hierarchy to Intelligence" in March 2026 and proposed something that sounds radical until you think about it for five minutes: a company built not as a machine but as an intelligence. A "mini-AGI," in their words. Not a hierarchy of humans routing information up and down a chain, but a living system that senses, thinks, and acts.

Jensen Huang told his CIO at NVIDIA that HR's future job is managing "digital employees" alongside biological ones — 7.5 million AI agents working with 75,000 humans. A hundred agents per person. That's not a machine. That's an ecosystem.

Satya Nadella calls it becoming "managers of infinite minds." Vinod Khosla says 80% of jobs will be done by AI by 2030 and $15 trillion of US labor GDP "will mostly go away." Sam Altman says a single person will run a billion-dollar company.

They're all describing the same thing from different angles. They're describing a company that is alive.

The anatomy

Think about what a living organism actually does. It has a brain that processes information and makes decisions. It has a nervous system that transmits signals and coordinates action. It has limbs that execute — that move, build, carry, manipulate the physical world. It has senses that perceive the environment. It has an immune system that detects threats and responds. It has memory. It metabolizes inputs into energy. It grows. It adapts. It heals.

Now map that onto the company that Dorsey is building at Block.

The brain is what Dorsey calls the "world model." Two AI systems: one that aggregates internal data — code, decisions, workflows, performance metrics — into a continuously updated picture of how the company operates. Another that maps external reality — customer behavior, merchant transactions, market signals. Together, they form the company's cognition. Not a CEO making decisions from a corner office. Not a management chain playing telephone with context. A shared intelligence layer that every part of the organism can access simultaneously.

This is the real revolution in the Dorsey-Botha essay. Hierarchy existed because information had to be routed. Someone at the bottom saw something. They told their manager. The manager told their director. The director told the VP. The VP told the CEO. By the time the signal reached the brain, it was distorted, delayed, and stripped of nuance. The company's "nervous system" was humans in a chain, and humans are lossy transmitters.

The world model replaces the chain with direct perception. Every node in the organism sees the same thing at the same time. The brain isn't a person. It's a system. And it doesn't sleep, forget, or have a bad day.

The nervous system is the coordination layer — the thing that connects perception to action. In the old model, this was middle management. Managers existed to translate strategy into tasks, to synchronize teams, to resolve conflicts between departments. They were the synapses of the corporate body, firing signals between the brain and the limbs.

Dorsey eliminated roughly 4,000 of those synapses in February 2026 — about 40% of Block's workforce. Not because the people were bad. Because the function they performed is now handled by the intelligence layer itself. When every IC has access to the world model, they don't need a manager to tell them what the company's priorities are. They can see for themselves. The nervous system went from biological to digital.

This is also what Nadella means by moving from the "brain phase" to the "nervous system phase" of AI deployment. The brain — the model — already exists. The hard part now is wiring it into everything, so that signals flow instantly from perception to decision to action without passing through five layers of human interpretation.

The limbs are the people and agents that execute. This is where Huang's 100-to-1 ratio matters. In a biological organism, the brain doesn't build the house. The hands do. The brain decides what to build and how, and the hands carry it out.

In the new company, the "hands" are a mix of human ICs and AI agents. Dorsey's framework has three human roles: individual contributors who build, directly responsible individuals who own outcomes for 90-day cycles, and player-coaches who develop others while still doing the work. Around them, AI agents handle the repetitive, the routine, the scalable. The organism doesn't distinguish between its biological and digital limbs. It uses whichever is better suited to the task.

This is how you get a one-person billion-dollar company. Not because one person does all the work. Because one person is the brain, and AI agents are the body. The organism is small in biological terms and enormous in functional terms. Sam Altman's prediction isn't about human capability. It's about the ratio of cognition to execution in a living system.

The senses are the data inputs — customer behavior, market signals, product telemetry, support tickets, sales calls, social media, competitive intelligence. In the old company, "sensing" was someone's job. A market researcher conducted surveys. A data analyst pulled reports. A product manager sat through customer calls. Each sense organ was a person, manually collecting and interpreting stimuli.

In the living company, sensing is continuous and ambient. The world model ingests everything in real-time. The organism doesn't wait for a quarterly business review to know what's happening. It knows what's happening the way you know when it's cold — automatically, without effort, because the nervous system is always on.

The immune system is governance, compliance, and quality control. Every organism needs the ability to detect and respond to threats — not just external threats (competitors, regulation, market shifts) but internal ones (bad decisions, misaligned agents, drift in quality). As companies deploy more AI agents, the immune function becomes critical. Who audits the agents? Who catches when an AI workflow starts producing subtly wrong outputs? Who ensures the organism doesn't harm itself?

This is the role that most companies haven't built yet, and it's the one that will matter most. A machine can run without an immune system — it just breaks and you fix it. An organism without an immune system dies.

Memory is institutional knowledge — the accumulated context of every decision, every customer interaction, every failure and success. In the old company, memory lived in people's heads. When they left, the memory left with them. In the living company, memory is externalized into the world model. The organism remembers everything, forever, and any part of it can access any memory instantly.

This is what makes the organism genuinely adaptive in a way that machines never were. A machine does what it was designed to do. An organism learns, remembers, and changes its behavior based on experience. A company with a world model that incorporates every past decision and its outcome can make better decisions tomorrow — not because a smart person is in charge, but because the system itself has learned.

Why the metaphor matters

This isn't just wordplay. The metaphor you use for a company determines how you build it.

If a company is a machine, you optimize for efficiency. You standardize processes. You make parts interchangeable. You minimize variance. You treat deviation as defect. This is Taylorism, and it works beautifully for predictable environments — factories, logistics, routine operations.

If a company is an organism, you optimize for adaptation. You build sensing capabilities. You distribute decision-making to the edges, because a nervous system that routes everything through a central brain is too slow. You invest in an immune system. You accept that the organism will sometimes do unexpected things, because that's what living systems do — and the unexpected is sometimes where the breakthrough comes from.

Dorsey's three roles — IC, DRI, player-coach — are the cellular structure of the organism. ICs are specialized cells that do specific work. DRIs are organs that own specific functions for specific periods. Player-coaches are stem cells that can differentiate into whatever the organism needs. There is no management layer because organisms don't have managers. They have nervous systems.

Andreessen's original thesis was "software is eating the world." By 2025, it had evolved to "AI is eating software." The organism metaphor takes it further: AI isn't eating the company. AI is the company. The biological components — the humans — are essential but increasingly outnumbered by the digital components. The question is no longer how to use AI within the company. The question is how humans fit within the AI organism.

The uncomfortable part

A living organism does something that machines don't. It sheds cells.

Your body replaces roughly 330 billion cells per day. It's not traumatic. It's maintenance. Old cells that no longer serve a function are discarded and replaced by new ones that do. The organism survives and thrives precisely because it doesn't hold onto what it doesn't need.

Block cut 4,000 people. The narrative was brutal — critics called the essay post-hoc justification for mass layoffs. And they're not entirely wrong to be uncomfortable. When you apply the organism metaphor to real humans with mortgages and families, the clinical language of "shedding cells" becomes obscene.

But the structural reality remains. If middle management existed to route information, and the intelligence layer now routes information, then the function has been replaced. The organism didn't fire people for the sake of it. The organism lost the need for a specific cell type. This will happen again. Khosla's 80% number may be aggressive on the timeline, but the direction isn't in dispute.

The question for any individual in the organism is not "is my job safe?" The question is "am I a cell the organism needs?" And that answer changes as the organism evolves. The cells that matter most in a living system aren't the ones that have been there the longest. They're the ones that are most responsive to what the organism needs right now.

Which brings us back to the roles that survive. In Dorsey's framework, the three roles that remain are all defined by judgment, not execution. ICs build — but they decide what to build, using the world model as context. DRIs own outcomes — but they decide which outcomes matter most for 90 days. Player-coaches develop people — but they decide how the organism's biological components need to grow. Every surviving role is a decision-making role. The organism automated everything else.

What the organism needs from you

Huang says every employee will manage 100 AI agents. Think about that in organism terms. You're not a single cell anymore. You're an organ — a functional unit that coordinates a cluster of digital cells toward a specific purpose. Your value isn't what you produce. It's what you direct.

Nadella's "managers of infinite minds" is the same insight. An infinite mind is not useful. An infinite mind pointed in the right direction by human judgment is transformative. The organism has unlimited capacity to act. What it needs from its biological components is the capacity to choose.

This is the new employment contract. The organism will give you leverage — AI agents, world model access, tools, compute. In return, you will give the organism judgment. You will sense things the digital nervous system misses. You will make calls that require empathy, ethics, creativity, taste — things the brain can't do alone. You will be the part of the organism that understands what it means to be human, because the organism's customers are human and its purpose is to serve them.

If that sounds abstract, make it concrete. Every morning you wake up as part of a living system. The system has already sensed the environment while you slept — it knows what customers are doing, what competitors launched, what broke overnight. Your job is to look at all of that and decide what matters. To say: this, not that. To allocate the organism's resources — your agents, your time, your attention — toward the thing that will help the organism survive and grow.

That's not a job description. That's a biological function.

The species is changing

We're watching a speciation event in real time. Some companies will remain machines — rigid, hierarchical, optimized for yesterday's environment. They'll function well in stable conditions and collapse when the environment shifts. Some companies will become organisms — adaptive, distributed, continuously learning. They'll be messy and unpredictable and sometimes they'll make mistakes that a well-designed machine never would. But they'll survive things that machines can't, because they can evolve.

The organism is not a better machine. It's a different category of thing. And the people who thrive inside it won't be the ones who were best at operating machines. They'll be the ones who understand what living systems need: sensing, judgment, adaptation, and the willingness to let go of what no longer serves the whole.

The company is alive. The question is whether you're a vital organ or a cell that's about to be replaced.

Act accordingly.