Tencent’s WorkBuddy team went from 10 people to 100 in under a year and got faster, not slower. What they actually redesigned was the operating model — decision distance, proximity to the real problem, and moving human judgment upstream into defining what “done” means.

Haier spent twenty years dismantling its own hierarchy. Block is betting AI can replace the coordination layer entirely. Opposite starting points, converging on the same question: what is middle management actually for once AI moves information through an organization?

Hierarchy was never mainly about power — it was a rational answer to information scarcity. AI just removed the scarcity. Which means self-management stops being idealism and becomes the economically rational design.

We have integrated extraordinary AI tools into our workflows while still measuring human value with performance systems built for a different era. Traditional models reward predictable execution and zero mistakes. AI-native talent runs ten bold experiments and fails eight times.

Five levels of how organisations actually scale AI agents, from zero to a thousand-plus. At every step your role shifts — and so does the bottleneck: from your attention, to your review capacity, to trust itself.

A CEO does not need to understand the technology — but they do need a tangible feel for what AI agents can actually deliver, or they cannot tell a real 20% efficiency gain from a well-packaged demo. Three things that cannot be delegated.

When agents do real work, headcount stops being the unit of capacity — and job architecture, performance management and workforce planning all rest on an assumption that no longer holds. One manufacturer’s answer: split HR into Human Service and a Human AI Resource Office.

An operating model for enterprises that need to absorb AI without breaking the people inside them. Five pillars — Distributed Authority, Fluid Structure, Evolutionary Purpose, Human Wholeness and Human–Agent Governance — and the three methods that make them practical.

Moonshot AI (Kimi) has about 300 people. By what it told its investors, its annual recurring revenue passed $1 billion in August 2026 — roughly $3 million per person. Midjourney: 11 people, $300 million. AI-native firms run at roughly $2M+ revenue per employee against about $250K in traditional software. That gap is not effort. It is structural.

Microsoft’s 2026 Work Trend Index found that 67% of AI adoption barriers are organizational, not individual. If your AI rollout has stalled, the constraint is almost certainly your approval layers, data access and incentives — not your people’s skills.

Tencent’s WorkBuddy team went from 10 people to 100 in under a year and got faster, not slower. What they actually redesigned was the operating model — decision distance, proximity to the real problem, and moving human judgment upstream into defining what “done” means.

Haier spent twenty years dismantling its own hierarchy. Block is betting AI can replace the coordination layer entirely. Opposite starting points, converging on the same question: what is middle management actually for once AI moves information through an organization?

Hierarchy was never mainly about power — it was a rational answer to information scarcity. AI just removed the scarcity. Which means self-management stops being idealism and becomes the economically rational design.

We have integrated extraordinary AI tools into our workflows while still measuring human value with performance systems built for a different era. Traditional models reward predictable execution and zero mistakes. AI-native talent runs ten bold experiments and fails eight times.

Five levels of how organisations actually scale AI agents, from zero to a thousand-plus. At every step your role shifts — and so does the bottleneck: from your attention, to your review capacity, to trust itself.

A CEO does not need to understand the technology — but they do need a tangible feel for what AI agents can actually deliver, or they cannot tell a real 20% efficiency gain from a well-packaged demo. Three things that cannot be delegated.

When agents do real work, headcount stops being the unit of capacity — and job architecture, performance management and workforce planning all rest on an assumption that no longer holds. One manufacturer’s answer: split HR into Human Service and a Human AI Resource Office.

An operating model for enterprises that need to absorb AI without breaking the people inside them. Five pillars — Distributed Authority, Fluid Structure, Evolutionary Purpose, Human Wholeness and Human–Agent Governance — and the three methods that make them practical.

Moonshot AI (Kimi) has about 300 people. By what it told its investors, its annual recurring revenue passed $1 billion in August 2026 — roughly $3 million per person. Midjourney: 11 people, $300 million. AI-native firms run at roughly $2M+ revenue per employee against about $250K in traditional software. That gap is not effort. It is structural.

Microsoft’s 2026 Work Trend Index found that 67% of AI adoption barriers are organizational, not individual. If your AI rollout has stalled, the constraint is almost certainly your approval layers, data access and incentives — not your people’s skills.

Tencent’s WorkBuddy team went from 10 people to 100 in under a year and got faster, not slower. What they actually redesigned was the operating model — decision distance, proximity to the real problem, and moving human judgment upstream into defining what “done” means.

Haier spent twenty years dismantling its own hierarchy. Block is betting AI can replace the coordination layer entirely. Opposite starting points, converging on the same question: what is middle management actually for once AI moves information through an organization?

Hierarchy was never mainly about power — it was a rational answer to information scarcity. AI just removed the scarcity. Which means self-management stops being idealism and becomes the economically rational design.

We have integrated extraordinary AI tools into our workflows while still measuring human value with performance systems built for a different era. Traditional models reward predictable execution and zero mistakes. AI-native talent runs ten bold experiments and fails eight times.

Five levels of how organisations actually scale AI agents, from zero to a thousand-plus. At every step your role shifts — and so does the bottleneck: from your attention, to your review capacity, to trust itself.

A CEO does not need to understand the technology — but they do need a tangible feel for what AI agents can actually deliver, or they cannot tell a real 20% efficiency gain from a well-packaged demo. Three things that cannot be delegated.

When agents do real work, headcount stops being the unit of capacity — and job architecture, performance management and workforce planning all rest on an assumption that no longer holds. One manufacturer’s answer: split HR into Human Service and a Human AI Resource Office.

An operating model for enterprises that need to absorb AI without breaking the people inside them. Five pillars — Distributed Authority, Fluid Structure, Evolutionary Purpose, Human Wholeness and Human–Agent Governance — and the three methods that make them practical.

Moonshot AI (Kimi) has about 300 people. By what it told its investors, its annual recurring revenue passed $1 billion in August 2026 — roughly $3 million per person. Midjourney: 11 people, $300 million. AI-native firms run at roughly $2M+ revenue per employee against about $250K in traditional software. That gap is not effort. It is structural.

Microsoft’s 2026 Work Trend Index found that 67% of AI adoption barriers are organizational, not individual. If your AI rollout has stalled, the constraint is almost certainly your approval layers, data access and incentives — not your people’s skills.

Tencent’s WorkBuddy team went from 10 people to 100 in under a year and got faster, not slower. What they actually redesigned was the operating model — decision distance, proximity to the real problem, and moving human judgment upstream into defining what “done” means.

Haier spent twenty years dismantling its own hierarchy. Block is betting AI can replace the coordination layer entirely. Opposite starting points, converging on the same question: what is middle management actually for once AI moves information through an organization?

Hierarchy was never mainly about power — it was a rational answer to information scarcity. AI just removed the scarcity. Which means self-management stops being idealism and becomes the economically rational design.

We have integrated extraordinary AI tools into our workflows while still measuring human value with performance systems built for a different era. Traditional models reward predictable execution and zero mistakes. AI-native talent runs ten bold experiments and fails eight times.

Five levels of how organisations actually scale AI agents, from zero to a thousand-plus. At every step your role shifts — and so does the bottleneck: from your attention, to your review capacity, to trust itself.

A CEO does not need to understand the technology — but they do need a tangible feel for what AI agents can actually deliver, or they cannot tell a real 20% efficiency gain from a well-packaged demo. Three things that cannot be delegated.

When agents do real work, headcount stops being the unit of capacity — and job architecture, performance management and workforce planning all rest on an assumption that no longer holds. One manufacturer’s answer: split HR into Human Service and a Human AI Resource Office.

An operating model for enterprises that need to absorb AI without breaking the people inside them. Five pillars — Distributed Authority, Fluid Structure, Evolutionary Purpose, Human Wholeness and Human–Agent Governance — and the three methods that make them practical.

Moonshot AI (Kimi) has about 300 people. By what it told its investors, its annual recurring revenue passed $1 billion in August 2026 — roughly $3 million per person. Midjourney: 11 people, $300 million. AI-native firms run at roughly $2M+ revenue per employee against about $250K in traditional software. That gap is not effort. It is structural.

Microsoft’s 2026 Work Trend Index found that 67% of AI adoption barriers are organizational, not individual. If your AI rollout has stalled, the constraint is almost certainly your approval layers, data access and incentives — not your people’s skills.

Tencent’s WorkBuddy team went from 10 people to 100 in under a year and got faster, not slower. What they actually redesigned was the operating model — decision distance, proximity to the real problem, and moving human judgment upstream into defining what “done” means.

Haier spent twenty years dismantling its own hierarchy. Block is betting AI can replace the coordination layer entirely. Opposite starting points, converging on the same question: what is middle management actually for once AI moves information through an organization?

Hierarchy was never mainly about power — it was a rational answer to information scarcity. AI just removed the scarcity. Which means self-management stops being idealism and becomes the economically rational design.

We have integrated extraordinary AI tools into our workflows while still measuring human value with performance systems built for a different era. Traditional models reward predictable execution and zero mistakes. AI-native talent runs ten bold experiments and fails eight times.

Five levels of how organisations actually scale AI agents, from zero to a thousand-plus. At every step your role shifts — and so does the bottleneck: from your attention, to your review capacity, to trust itself.

A CEO does not need to understand the technology — but they do need a tangible feel for what AI agents can actually deliver, or they cannot tell a real 20% efficiency gain from a well-packaged demo. Three things that cannot be delegated.

When agents do real work, headcount stops being the unit of capacity — and job architecture, performance management and workforce planning all rest on an assumption that no longer holds. One manufacturer’s answer: split HR into Human Service and a Human AI Resource Office.

An operating model for enterprises that need to absorb AI without breaking the people inside them. Five pillars — Distributed Authority, Fluid Structure, Evolutionary Purpose, Human Wholeness and Human–Agent Governance — and the three methods that make them practical.

Moonshot AI (Kimi) has about 300 people. By what it told its investors, its annual recurring revenue passed $1 billion in August 2026 — roughly $3 million per person. Midjourney: 11 people, $300 million. AI-native firms run at roughly $2M+ revenue per employee against about $250K in traditional software. That gap is not effort. It is structural.

Microsoft’s 2026 Work Trend Index found that 67% of AI adoption barriers are organizational, not individual. If your AI rollout has stalled, the constraint is almost certainly your approval layers, data access and incentives — not your people’s skills.

Tencent’s WorkBuddy team went from 10 people to 100 in under a year and got faster, not slower. What they actually redesigned was the operating model — decision distance, proximity to the real problem, and moving human judgment upstream into defining what “done” means.

Haier spent twenty years dismantling its own hierarchy. Block is betting AI can replace the coordination layer entirely. Opposite starting points, converging on the same question: what is middle management actually for once AI moves information through an organization?

Hierarchy was never mainly about power — it was a rational answer to information scarcity. AI just removed the scarcity. Which means self-management stops being idealism and becomes the economically rational design.

We have integrated extraordinary AI tools into our workflows while still measuring human value with performance systems built for a different era. Traditional models reward predictable execution and zero mistakes. AI-native talent runs ten bold experiments and fails eight times.

Five levels of how organisations actually scale AI agents, from zero to a thousand-plus. At every step your role shifts — and so does the bottleneck: from your attention, to your review capacity, to trust itself.

A CEO does not need to understand the technology — but they do need a tangible feel for what AI agents can actually deliver, or they cannot tell a real 20% efficiency gain from a well-packaged demo. Three things that cannot be delegated.

When agents do real work, headcount stops being the unit of capacity — and job architecture, performance management and workforce planning all rest on an assumption that no longer holds. One manufacturer’s answer: split HR into Human Service and a Human AI Resource Office.

An operating model for enterprises that need to absorb AI without breaking the people inside them. Five pillars — Distributed Authority, Fluid Structure, Evolutionary Purpose, Human Wholeness and Human–Agent Governance — and the three methods that make them practical.

Moonshot AI (Kimi) has about 300 people. By what it told its investors, its annual recurring revenue passed $1 billion in August 2026 — roughly $3 million per person. Midjourney: 11 people, $300 million. AI-native firms run at roughly $2M+ revenue per employee against about $250K in traditional software. That gap is not effort. It is structural.

Microsoft’s 2026 Work Trend Index found that 67% of AI adoption barriers are organizational, not individual. If your AI rollout has stalled, the constraint is almost certainly your approval layers, data access and incentives — not your people’s skills.

Tencent’s WorkBuddy team went from 10 people to 100 in under a year and got faster, not slower. What they actually redesigned was the operating model — decision distance, proximity to the real problem, and moving human judgment upstream into defining what “done” means.

Haier spent twenty years dismantling its own hierarchy. Block is betting AI can replace the coordination layer entirely. Opposite starting points, converging on the same question: what is middle management actually for once AI moves information through an organization?

Hierarchy was never mainly about power — it was a rational answer to information scarcity. AI just removed the scarcity. Which means self-management stops being idealism and becomes the economically rational design.

We have integrated extraordinary AI tools into our workflows while still measuring human value with performance systems built for a different era. Traditional models reward predictable execution and zero mistakes. AI-native talent runs ten bold experiments and fails eight times.

Five levels of how organisations actually scale AI agents, from zero to a thousand-plus. At every step your role shifts — and so does the bottleneck: from your attention, to your review capacity, to trust itself.

A CEO does not need to understand the technology — but they do need a tangible feel for what AI agents can actually deliver, or they cannot tell a real 20% efficiency gain from a well-packaged demo. Three things that cannot be delegated.

When agents do real work, headcount stops being the unit of capacity — and job architecture, performance management and workforce planning all rest on an assumption that no longer holds. One manufacturer’s answer: split HR into Human Service and a Human AI Resource Office.

An operating model for enterprises that need to absorb AI without breaking the people inside them. Five pillars — Distributed Authority, Fluid Structure, Evolutionary Purpose, Human Wholeness and Human–Agent Governance — and the three methods that make them practical.

Moonshot AI (Kimi) has about 300 people. By what it told its investors, its annual recurring revenue passed $1 billion in August 2026 — roughly $3 million per person. Midjourney: 11 people, $300 million. AI-native firms run at roughly $2M+ revenue per employee against about $250K in traditional software. That gap is not effort. It is structural.

Microsoft’s 2026 Work Trend Index found that 67% of AI adoption barriers are organizational, not individual. If your AI rollout has stalled, the constraint is almost certainly your approval layers, data access and incentives — not your people’s skills.

Tencent’s WorkBuddy team went from 10 people to 100 in under a year and got faster, not slower. What they actually redesigned was the operating model — decision distance, proximity to the real problem, and moving human judgment upstream into defining what “done” means.

Haier spent twenty years dismantling its own hierarchy. Block is betting AI can replace the coordination layer entirely. Opposite starting points, converging on the same question: what is middle management actually for once AI moves information through an organization?

Hierarchy was never mainly about power — it was a rational answer to information scarcity. AI just removed the scarcity. Which means self-management stops being idealism and becomes the economically rational design.

We have integrated extraordinary AI tools into our workflows while still measuring human value with performance systems built for a different era. Traditional models reward predictable execution and zero mistakes. AI-native talent runs ten bold experiments and fails eight times.

Five levels of how organisations actually scale AI agents, from zero to a thousand-plus. At every step your role shifts — and so does the bottleneck: from your attention, to your review capacity, to trust itself.

A CEO does not need to understand the technology — but they do need a tangible feel for what AI agents can actually deliver, or they cannot tell a real 20% efficiency gain from a well-packaged demo. Three things that cannot be delegated.

When agents do real work, headcount stops being the unit of capacity — and job architecture, performance management and workforce planning all rest on an assumption that no longer holds. One manufacturer’s answer: split HR into Human Service and a Human AI Resource Office.

An operating model for enterprises that need to absorb AI without breaking the people inside them. Five pillars — Distributed Authority, Fluid Structure, Evolutionary Purpose, Human Wholeness and Human–Agent Governance — and the three methods that make them practical.

Moonshot AI (Kimi) has about 300 people. By what it told its investors, its annual recurring revenue passed $1 billion in August 2026 — roughly $3 million per person. Midjourney: 11 people, $300 million. AI-native firms run at roughly $2M+ revenue per employee against about $250K in traditional software. That gap is not effort. It is structural.

Microsoft’s 2026 Work Trend Index found that 67% of AI adoption barriers are organizational, not individual. If your AI rollout has stalled, the constraint is almost certainly your approval layers, data access and incentives — not your people’s skills.

Tencent’s WorkBuddy team went from 10 people to 100 in under a year and got faster, not slower. What they actually redesigned was the operating model — decision distance, proximity to the real problem, and moving human judgment upstream into defining what “done” means.

Haier spent twenty years dismantling its own hierarchy. Block is betting AI can replace the coordination layer entirely. Opposite starting points, converging on the same question: what is middle management actually for once AI moves information through an organization?

Hierarchy was never mainly about power — it was a rational answer to information scarcity. AI just removed the scarcity. Which means self-management stops being idealism and becomes the economically rational design.

We have integrated extraordinary AI tools into our workflows while still measuring human value with performance systems built for a different era. Traditional models reward predictable execution and zero mistakes. AI-native talent runs ten bold experiments and fails eight times.

Five levels of how organisations actually scale AI agents, from zero to a thousand-plus. At every step your role shifts — and so does the bottleneck: from your attention, to your review capacity, to trust itself.

A CEO does not need to understand the technology — but they do need a tangible feel for what AI agents can actually deliver, or they cannot tell a real 20% efficiency gain from a well-packaged demo. Three things that cannot be delegated.

When agents do real work, headcount stops being the unit of capacity — and job architecture, performance management and workforce planning all rest on an assumption that no longer holds. One manufacturer’s answer: split HR into Human Service and a Human AI Resource Office.

An operating model for enterprises that need to absorb AI without breaking the people inside them. Five pillars — Distributed Authority, Fluid Structure, Evolutionary Purpose, Human Wholeness and Human–Agent Governance — and the three methods that make them practical.

Moonshot AI (Kimi) has about 300 people. By what it told its investors, its annual recurring revenue passed $1 billion in August 2026 — roughly $3 million per person. Midjourney: 11 people, $300 million. AI-native firms run at roughly $2M+ revenue per employee against about $250K in traditional software. That gap is not effort. It is structural.

Microsoft’s 2026 Work Trend Index found that 67% of AI adoption barriers are organizational, not individual. If your AI rollout has stalled, the constraint is almost certainly your approval layers, data access and incentives — not your people’s skills.

Tencent’s WorkBuddy team went from 10 people to 100 in under a year and got faster, not slower. What they actually redesigned was the operating model — decision distance, proximity to the real problem, and moving human judgment upstream into defining what “done” means.

Haier spent twenty years dismantling its own hierarchy. Block is betting AI can replace the coordination layer entirely. Opposite starting points, converging on the same question: what is middle management actually for once AI moves information through an organization?

Hierarchy was never mainly about power — it was a rational answer to information scarcity. AI just removed the scarcity. Which means self-management stops being idealism and becomes the economically rational design.

We have integrated extraordinary AI tools into our workflows while still measuring human value with performance systems built for a different era. Traditional models reward predictable execution and zero mistakes. AI-native talent runs ten bold experiments and fails eight times.

Five levels of how organisations actually scale AI agents, from zero to a thousand-plus. At every step your role shifts — and so does the bottleneck: from your attention, to your review capacity, to trust itself.

A CEO does not need to understand the technology — but they do need a tangible feel for what AI agents can actually deliver, or they cannot tell a real 20% efficiency gain from a well-packaged demo. Three things that cannot be delegated.

When agents do real work, headcount stops being the unit of capacity — and job architecture, performance management and workforce planning all rest on an assumption that no longer holds. One manufacturer’s answer: split HR into Human Service and a Human AI Resource Office.

An operating model for enterprises that need to absorb AI without breaking the people inside them. Five pillars — Distributed Authority, Fluid Structure, Evolutionary Purpose, Human Wholeness and Human–Agent Governance — and the three methods that make them practical.

Moonshot AI (Kimi) has about 300 people. By what it told its investors, its annual recurring revenue passed $1 billion in August 2026 — roughly $3 million per person. Midjourney: 11 people, $300 million. AI-native firms run at roughly $2M+ revenue per employee against about $250K in traditional software. That gap is not effort. It is structural.

Microsoft’s 2026 Work Trend Index found that 67% of AI adoption barriers are organizational, not individual. If your AI rollout has stalled, the constraint is almost certainly your approval layers, data access and incentives — not your people’s skills.