Notion CEO Talks AI Transformation: The Era of "Infinite Intelligence" Has Arrived

Recently, Notion co-founder and CEO Ivan Zhao published an in-depth article titled "Steam, Steel, and Infinite Minds" on his official blog.

In the tech hub of San Francisco, despite the ongoing discussions about Artificial General Intelligence (AGI), Ivan Zhao pointed out that billions of knowledge workers around the world have yet to truly feel its impact. Through historical metaphors like steel and steam engines, he profoundly analyzed how AI is reshaping individuals, organizations, and even entire economies.

Ivan Zhao believes we are in a painful period of technological transformation, stating, "The most popular form of artificial intelligence today is akin to the old Google search box," and that people are, as Marshall McLuhan said, "always driving into the future looking through the rearview mirror."

At the individual level, the transformation is already evident among programmers. Ivan Zhao used his partner Simon as an example, describing the leap from "thought bicycles" to "cars": "Passing by his workstation, you would see him simultaneously directing three or four AI programming agents. These AIs not only input faster but also possess thinking capabilities... He has become a manager of infinite minds."

At the organizational level, Ivan Zhao likened AI to "the steel of organizations." Just as steel allowed skyscrapers to surpass the limitations of brick and wood structures, AI will break the bottlenecks of organizational scale. "AI is the steel of organizations... Human communication no longer has to bear the weight: a two-hour weekly meeting can be compressed into a five-minute asynchronous review... Companies will achieve truly lossless scale expansion." Meanwhile, the industry is still in the "waterwheel era," simply grafting AI onto old processes without thoroughly reconstructing production processes like in the steam engine era.

At the economic level, Ivan Zhao predicted that the knowledge economy will undergo a transformation from "Florence" to "super metropolis." Existing organizations are like "building Florence with stones and wood," limited by human scales; AI will construct "Tokyo-style" organizations—"a collaborative network of thousands of agents and humans, operating continuously across time zones." This change, while bringing about "incomprehensibility" and a sense of disorientation, will yield unprecedented scale and speed.

At the end of the article, Ivan Zhao revealed the experimental progress within Notion: "In addition to 1,000 employees, there are now over 700 agents handling repetitive tasks... and this is just the starting phase." He urged the industry to stop thinking in the rearview mirror, "Steel. Steam. Infinite minds. The next skyline is right there, waiting for us to build."

Productivity software company Notion is a super unicorn headquartered in San Francisco, which has been striving to position itself as the "universal application" in the office domain, challenging Microsoft and Google’s dominance in the productivity suite market. Notion continuously rolls out new AI features aimed at creating an integrated office platform, providing users with comprehensive solutions from note-taking to knowledge management Core Points Summary:

• From "Cycling" to "Driving": AI agents upgrade knowledge workers from "thought cyclists" to "infinite minds" managers, akin to transitioning from riding a bicycle to driving a car.
• Farewell to the "Waterwheel Era": Current applications are still in the phase of simply grafting chatbots onto old processes; true transformation lies in reconstructing workflows around AI, just as factories transitioned from being water-powered to steam-powered.
• The "Steel" of Organizations: AI is the "steel" of modern organizations, breaking the weight limit of human communication and enabling enterprises to achieve true lossless scalable expansion.
• From Florence to Tokyo: The economies of the AI era will evolve from the "human-scale" Florence model to the high-density, high-speed, round-the-clock operating "Tokyo-style" super city model.
• Notion's Practice: Notion already has over 700 agents collaborating with 1,000 employees to handle repetitive tasks, marking just the beginning of the "infinite mind" era.

Full Translation:

Steam, Steel, and Infinite Minds

Author: Ivan Zhao, Co-founder and CEO Published on: December 23, 2025

Every era is shaped by its "miracle materials." Steel forged the gilded age. Semiconductors ushered in the digital age. And now, AI has arrived in the form of "infinite minds." If history has taught us anything, it is that those who master the core materials will ultimately define the era.

In the 1850s, Andrew Carnegie was a telegraph operator on the muddy streets of Pittsburgh. At that time, sixty percent of Americans were farmers. But within just two generations, Carnegie and his peers forged the modern world. Railroads replaced horse-drawn carriages, electric lights replaced candles, and steel revolutionized pig iron.

Since then, the focus of work has shifted from factories to offices. Today, I run a software company in San Francisco, building tools for millions of knowledge workers. In this tech hub, everyone is talking about Artificial General Intelligence (AGI), yet most of the two billion desk workers worldwide have yet to truly feel its impact. What will knowledge work look like in the near future? What will happen when organizational structures integrate an ever-awake mind?

The future is often hard to predict because it always disguises itself as the past. Early telephone calls were as concise as telegrams. Early films looked like recorded stage plays. (This is what Marshall McLuhan meant by "driving into the future through the rearview mirror.") The most popular form of artificial intelligence today resembles the old Google search box. To quote Marshall McLuhan: "We drive into the future using only our rearview mirror."

Today, we see AI chatbots mimicking the Google search box. We are deeply entrenched in that unsettling transition period that accompanies every new technological iteration.

I do not have all the answers about what will happen next. But I like to use some historical metaphors to think about how AI operates on different scales, from individuals to organizations, and then to entire economies.

Individual Level: From Bicycle to Car

The initial signs of transformation are evident in the "priestly class" of knowledge work: the programmer community.

My partner Simon was what we call a "10x programmer," but nowadays he rarely writes code himself. Passing by his workstation, you would see him simultaneously directing three or four AI programming agents. These agents not only input faster, but they also possess greater cognitive abilities, making him a "30-40x efficiency engineer." He can assign tasks during lunch breaks or before bed, allowing the agents to continue working while he is away. He has become a manager of infinite intellect.

A study on athletic performance published in Scientific American in the 1970s inspired Steve Jobs' famous metaphor of the "bicycle of the mind." However, for decades since, we have been pedaling bicycles on the information superhighway.

In the 1980s, Steve Jobs referred to personal computers as the "bicycle of the mind." A decade later, we paved the "information superhighway" with the internet. Yet today, most knowledge work still relies on human power. It's like riding a bicycle on the highway.

With AI agents, people like Simon have completed the upgrade from riding a bicycle to driving a car.

When will other types of knowledge workers be able to drive cars? Two challenges must be addressed.

Why is it more difficult for AI to assist general knowledge work compared to programming agents? Because the former faces issues of fragmented scenarios and difficult-to-verify outcomes.

First, there is Context Fragmentation. In programming, tools and context are often centralized: IDEs, code repositories, terminals. However, general knowledge work is scattered across dozens of tools. Imagine an AI agent trying to draft a product brief: it needs to extract information from Slack discussion threads, strategic documents, last quarter's data dashboards, and organizational memory that exists only in someone's mind. Today, humans still act as the glue, piecing together all the information through copy-pasting and switching between browser tabs. Unless these contexts are integrated, the agent will be trapped in narrow application scenarios.

The second missing element is Verifiability. Code has a magical property: you can verify it through testing and error reporting. Model developers leverage this to train AI to be better at programming (e.g., reinforcement learning). But how do you verify whether a project is managed well or whether a strategic memo is excellent? We have yet to find ways to improve the models of general knowledge work. Therefore, humans still need to be "in the loop" for supervision, guidance, and demonstrating what good standards are.

The Red Flag Act of 1865 required a person to walk in front of vehicles holding a red flag while they were in motion (the act was repealed in 1896). This is an unpopular example of "humans in the loop."

This year's practice with programming agents has taught us that "humans in the loop" is not always ideal. It's like having someone check every screw on a production line or requiring a flag bearer to walk in front of cars (see the 1865 Red Flag Act). We want humans to supervise from a leverage point rather than being stuck in the loop. Once context is integrated and work can be verified, billions of workers will upgrade from pedaling bicycles to driving cars, ultimately moving towards autonomous driving.

Organizational Level: Steel and Steam

Companies are a modern invention. As they scale, they experience diminishing returns and hit limits.

Hundreds of years ago, most companies were just workshops with a dozen people. Today, we have multinational corporations with hundreds of thousands of employees. The communication infrastructure (the human brains connected through meetings and information) is overwhelmed under exponential loads. We try to solve this dilemma with hierarchies, processes, and documentation. But we have been using tools designed for human scales to solve industrial-level problems, like building skyscrapers with wood.

Two historical metaphors reveal how novel materials will reshape future organizations.

The first is steel. Before the advent of steel, buildings in the 19th century could only be constructed up to six or seven stories. Iron, while strong, was brittle and heavy; the more floors added, the more likely the structure would collapse under its own weight. Steel changed everything. It is both strong and ductile. Frames can be lighter, walls can be thinner, and suddenly, buildings can be constructed dozens of stories high. New types of architecture became possible.

AI is the steel of organizations. It has the potential to maintain contextual awareness in workflows and trigger decisions accurately when needed without information overload. Human communication no longer needs to act as a load-bearing wall. Weekly two-hour alignment meetings can turn into five-minute asynchronous reviews. Executive decisions that previously required three levels of approval could soon be completed in minutes. Companies will achieve scalability, true scalability, without enduring the efficiency decay we once viewed as inevitable.

The second story is about the steam engine. In the early days of the Industrial Revolution, early textile mills were built along rivers and streams, powered by waterwheels. When the steam engine appeared, mill owners initially just replaced the waterwheel with a steam engine, keeping everything else the same. The productivity gains were quite limited.

The real breakthrough occurred when mill owners realized they could completely free themselves from the constraints of water power. They built larger factories closer to workers, ports, and raw materials. Moreover, they redesigned factories around the steam engine (later, when electricity became widespread, mill owners further abandoned central power shafts, distributing small motors throughout the factory to drive different machines). Productivity then exploded, and the Second Industrial Revolution truly took off.

We are still in the "replacing the waterwheel" phase. Simply grafting AI chatbots onto existing tools. When the old constraints crumble, and your company can rely on an ever-awake, infinite intellect, we have yet to reimagine what organizations will look like.

At my company Notion, we have been experimenting. In addition to 1,000 employees, there are now over 700 agents responsible for handling repetitive tasks. They take meeting minutes and answer questions to consolidate tribal knowledge. They handle IT requests and record customer feedback. They help new employees understand employee benefits. They write weekly status reports so people don’t have to copy and paste. And this is just the starting phase. The real benefits are only limited by our imagination and inertia.

Economic Aspect: From Florence to Megacities

Steel and steam not only changed buildings and factories. They changed cities

Until a few hundred years ago, cities were built on a human scale. You could walk across Florence in forty minutes. The rhythm of life depended on how far a person could walk and how loudly sounds could carry.

Later, steel frame structures made skyscrapers possible. Steam engines powered the railways connecting city centers to the hinterlands. Elevators, subways, and highways followed in quick succession. Cities exploded in scale and density. Tokyo. Chongqing. Dallas.

These are not just enlarged versions of Florence. They represent entirely different ways of living. Supercities can be disorienting, filled with anonymity, and difficult to navigate. This "illegibility" is the cost of scale. But they also offer more opportunities and more freedom, accommodating a far greater combination of human activities than the cities of the Renaissance could support.

I believe the knowledge economy is about to undergo a similar transformation.

Today, knowledge work accounts for nearly half of the U.S. GDP. But most of it still operates on a human scale: teams of dozens, workflows set by meetings and emails, organizations that hit bottlenecks once they exceed a few hundred people. We have built our "Florence" with stone and wood.

When AI agents scale up, we will build our "Tokyo." A collaborative network accommodating thousands of agents and humans. Workflows running continuously across time zones, without waiting for someone to wake up. Decision-making mechanisms precisely embedding just the right amount of humans in the loop.

This will bring a completely new experience. Faster, more leveraged, but initially also more confusing. The rhythms of weekly meetings, quarterly planning, and annual evaluations may no longer apply. A new rhythm is about to be born. We will lose some readability. We will gain scale and speed.

Beyond the Waterwheel Era

Every miraculous 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. The mill owners of Lancashire looked at the steam engine and saw factories freed from the constraints of rivers.

We are still in the waterwheel era of AI, merely attaching chatbots to workflows designed for humans. We should no longer just ask AI to be our co-pilot. We need to imagine what knowledge work will look like when human organizations are reinforced with the strength of steel, and tedious tasks are delegated to minds that never sleep.

Steel. Steam. Infinite minds. The next skyline is out there, waiting for us to build it