Notion CEO discusses the AI revolution: The era of "infinite mind" has arrived.
```
Recently, Ivan Zhao, co-founder and CEO of Notion, published an in-depth article titled "Steam, Steel, and Infinite Minds" on the official blog.
In the tech hub of San Francisco, although discussions about Artificial General Intelligence (AGI) are ever-present, Ivan Zhao points out that billions of knowledge workers worldwide have yet to truly feel its impact. Using historical metaphors such as steel and the steam engine, he deeply analyzes how AI is reshaping individuals, organizations, and even the entire economy.
Ivan Zhao believes that we are in the painful transition of a technological shift. "The most popular form of AI today is like the Google search box of the past." People, as Marshall McLuhan said, "are always driving toward the future through the rearview mirror."
At the individual level, change has already manifested within the programmer community. Ivan Zhao uses his partner Simon as an example to describe the leap from “bicycle for the mind” to “car”: “Passing by his desk, you’ll see him commanding three or four AI programming agents at once. These AIs are not only faster at input, but also capable of thinking… He has become a manager of infinite minds.”
At the organizational level, Ivan Zhao compares AI to “the steel of organizations.” Just as steel allowed skyscrapers to break through the limitations of brick and wood structures, AI will break the bottleneck of organizational scale. “AI is the steel of organizations… Human communication no longer needs to bear the load: a two-hour weekly meeting can be compressed into five minutes of asynchronous recap… Companies will achieve truly lossless scaling." At the same time, the industry is still in the "waterwheel era," only grafting AI onto old processes, rather than thoroughly reconstructing workflows as happened in the steam engine era.
At the economic level, Ivan Zhao predicts the knowledge economy will transform from “Florence” to “megacity.” Existing organizations are like "Florence built with stone and wood," limited by the scale of human labor, while AI will build "Tokyo-style" organizations — "collaborative networks of thousands of agents and humans, with workflows operating nonstop across time zones." Although this change will bring "illegibility" and a sense of being lost, it will bring unprecedented scale and speed.
At the end of the article, Ivan Zhao reveals some experimental progress within Notion: "In addition to 1,000 employees, there are now more than 700 agents handling repetitive work... and this is just the starting stage." He calls on the industry to stop rear-view mirror thinking: "Steel. Steam. Infinite minds. The next horizon is already there, waiting for us to build it."
Notion, a productivity software company, is a super-unicorn headquartered in San Francisco. In recent years, it has been striving to become an "all-in-one workspace" in the office domain, challenging Microsoft and Google’s dominance in productivity suites. Notion continuously launches new AI features, aiming to create an integrated office platform that provides users with comprehensive solutions from note-taking to knowledge management.
Key Points:
From “Cycling” to “Driving”: AI agents upgrade knowledge workers from “cycling on a bicycle for the mind” to “managers of infinite minds,” like switching from riding a bike to driving a car.Farewell to the “Waterwheel Era”: Current applications are still at the stage of simply grafting chatbots onto old workflows. The real transformation is to reconstruct workflows around AI, just like factories turned from water-powered to steam-powered.“Steel” for Organizations: AI is the “steel” for modern organizations; it can break the upper limit of human communication load, enabling truly lossless organizational scaling.From Florence to Tokyo: The AI era economy will evolve from a “Florence model” built at human scale into a “Tokyo-style” megacity — high-density, high-speed, and operating around the clock.Notion’s Practice: Notion already has more than 700 agents collaborating with 1,000 employees to handle repetitive tasks — only the beginning of the “infinite minds” era.
Full Translation:
Steam, Steel, and Infinite Minds
By Ivan Zhao, Co-founder and CEO, Published: December 23, 2025
Every era is shaped by its “miracle material.” Steel forged the Gilded Age. Semiconductors ushered in the Digital Era. Now, AI has arrived in the form of “infinite minds.” If history teaches us anything, it’s that those who grasp the core material will inevitably define the age.
In the 1850s, Andrew Carnegie was a telegraph operator on the muddy streets of Pittsburgh. At that time, 60% of Americans were farmers. But in just two generations, Carnegie and his peers forged the modern world. Railroads replaced horse-drawn carriages, electric lamps replaced candles, and steel revolutionized pig iron.
Ever since, the center of work has shifted from the factory to the office. Today I run a software company in San Francisco, building tools for millions of knowledge workers. In this tech hub, everyone talks about Artificial General Intelligence (AGI), but the vast majority of the world’s two billion desk workers have yet to actually feel its impact. What will knowledge work become in the near future? What happens when organizational structure is fused with unceasing minds?
The future is often hard to predict, because it always masquerades as the past. Early telephone calls were as terse as telegrams. Early movies looked like recorded stage plays. (This is what Marshall McLuhan called “driving into the future through the rearview mirror.”)
The most popular form of AI today looks very much like the Google search box of yesteryear. Borrowing McLuhan’s phrase: “We are always driving toward the future through the rearview mirror.”
Today, we see AI chatbots mimicking the Google search box. We are mired in that discomforting transition that every technological turnover brings.
I don’t have all the answers about what comes next. But I like to use historical metaphors to think about how AI operates at different scales: from individuals to organizations to the entire economy.
Individual Level: From Bicycle to Car
The first signs of change appear among the "priesthood" of knowledge work: programmers.
My partner Simon used to be what we called a “10x engineer,” but now he rarely writes code himself. Passing by his desk, you’ll see him commanding three or four AI programming agents at once. These agents are not only faster at input, they also have the ability to think, which makes him a "30-40x engineer." He can assign tasks before lunch or bed, and the agents keep working while he’s gone. He’s become a manager of infinite minds.
A 1970s Scientific American study of locomotion efficiency inspired Steve Jobs’ famous “bicycle for the mind” metaphor. But in the decades since, we have always been pedaling on the information highway.
In the 1980s, Steve Jobs called the personal computer a “bicycle for the mind.” A decade later, we laid out the “information highway” of the internet. Yet today, most knowledge work still depends on human muscle. It’s like we’re riding bicycles on a superhighway.
With AI agents, people like Simon have upgraded from cycling to driving a car.
When will other kinds of knowledge workers get to drive cars? Two things must be solved.
Why is it harder for AI to help with general knowledge work than with programming? Because the former has fragmented contexts and results that are hard to verify.
First is context fragmentation. For programming, tools and context are often in one place: the IDE, codebase, terminal. But general knowledge work is scattered across dozens of tools. Imagine an agent trying to draft a product brief: it needs to pull from Slack threads, strategy documents, last quarter’s dashboards, and organization memory that only exists in someone’s head. Today, humans are the glue, copy-pasting and switching between browser tabs to piece this together. Unless these contexts are unified, agents will be trapped in narrow applications.
The second missing piece is verifiability. Code has a magical property: you can verify it with tests and errors. Model trainers use this to teach AIs to code (e.g., reinforcement learning). But how do you verify if a project is well managed or a strategic memo is excellent? We haven’t found ways to improve models for general knowledge work. So humans are still "in the loop" — supervising, guiding, and showing what good standards look like.
The 1865 “Red Flag Act” required vehicles to be preceded by a man carrying a red flag (the law was repealed in 1896). This was an unpopular “human in the loop” example.
This year’s agent experiments in programming taught us that “human in the loop” isn’t always ideal. It’s like arranging for someone to check every screw on an assembly line or having a red flag bearer walk ahead of every car (see the Red Flag Act of 1865). We want humans supervising from the fulcrum — not stuck in the loop. Once context is integrated and work can be verified, billions of workers will upgrade from cycling to driving, and ultimately to autonomous driving.
Organizational Level: Steel and Steam
The company is a modern invention. As companies grew, their efficiency waned and limits were reached.
Centuries ago, most companies were workshops of a dozen people. Now we have multinationals with hundreds of thousands of employees. Communication infrastructure (humans linking brains in meetings and messages) is overwhelmed by exponential loads. We try to solve this with hierarchies, procedures, and documentation. But we’re using human-scale tools to solve industrial-scale problems — building skyscrapers out of wood.
Two historical metaphors show how the new miracle material will reshape organizations.
First: Steel. Before steel, 19th-century buildings could only be six or seven stories high. Iron was strong, but brittle and heavy; add more floors and the structure would collapse under its weight. Steel changed everything. It was strong and ductile. Frames could be lighter, walls thinner, and suddenly buildings could soar dozens of stories. New forms became possible.
AI is the steel of organizations. It can maintain context-sensitive workflows and trigger decisions precisely without overload. Human communication need not bear the weight. Two-hour alignment meetings can become five-minute asynchronous recaps. Executive decisions that needed three levels of approval can soon be made in minutes. Companies will scale — truly scale — without suffering the efficiency loss we once thought inevitable.
The second story is about the steam engine. In the early Industrial Revolution, textile mills were built by rivers and powered by waterwheels. When steam engines arrived, mill owners at first just replaced waterwheels with steam engines, keeping everything else the same. Productivity barely improved.
The real breakthrough came when owners realized they could be free from water constraints. They built larger factories closer to workers, ports, and raw materials. They redesigned the factories around the steam engine (and, later, as electricity spread, stopped using central power shafts and put electric motors at different points in the factory). Productivity soared, and the true Second Industrial Revolution took off.
We are still at the "replacing the waterwheel" stage — just grafting chatbots onto legacy workflows. When old constraints fall away, and your company can run on infinite, sleepless minds, we still haven’t reimagined what organizations could be.
At Notion, we are experimenting. In addition to 1,000 employees, there are now more than 700 agents handling repetitive work. They record meeting notes, answer questions to unify tribal knowledge. They process IT tickets and record customer feedback. They help new hires understand benefits. They write weekly status reports so people don’t need to copy-paste. And this is just the beginning. The true benefits are limited only by our imagination and inertia.
Economic Level: From Florence to Megacity
Steel and steam did not just change buildings and factories. They changed the city.
Until a few centuries ago, cities were built to human scale — you could walk across Florence in forty minutes. The pace of life depended on how far a person could walk and how loud a voice could carry.
Then steel frames made skyscrapers possible. Steam powered the trains that connected city centers to their hinterlands. Elevators, subways, expressways followed. Cities burst in scale and density. Tokyo. Chongqing. Dallas.
These are not just scaled-up Florences. They are different ways of living. Megacities are bewildering, anonymous, hard to navigate. This “illegibility” is the price of scale. But they also offer more opportunity, more freedom — combinations of human activity far exceeding anything Renaissance towns could contain.
I believe the knowledge economy is about to undergo the same transformation.
Today, knowledge work is nearly half of US GDP. But most of it still operates at human scale: teams of a few dozen, workflows paced by meetings and emails, organizations bottlenecked above a few hundred people. We built “Florence” out of stone and wood.
With AI agents scaled up, we’ll build “Tokyo”: collaborative networks of thousands of agents and people, with 24/7 workflows running across time zones, no more waiting for someone to wake up. Decision-making embeds just the right amount of human in the loop.
This will be a new kind of experience: faster, more leveraged, but at first, also more bewildering. Weekly meetings, quarterly plans, annual reviews might not fit anymore. A new rhythm is about to be born. We’ll lose some legibility; we’ll gain scale and speed.
Beyond the Waterwheel Era
Every miracle material demands that people stop looking at the world through the rearview mirror and start imagining a new world. Carnegie looked at steel and saw the city skyline. Lancashire mill owners saw steam engines and imagined factories untethered from rivers.
We’re still in AI’s waterwheel era, just bolting chatbots onto workflows built for humans. We shouldn’t just ask AI to be our co-pilot. We should imagine what knowledge work will be like when human organizations get steel-like enhancement, and tedious work is delegated to infinite, sleepless minds.
Steel. Steam. Infinite minds. The next skyline is already there, waiting for us to build it.
Risk warnings and disclaimersThe market has risks; investment must be cautious. This article does not constitute personal investment advice, nor does it take into account the special investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Those who invest accordingly do so at their own risk. ```