Originally published on LinkedIn on July 6, 2026.
"We spent the last century learning how to make organizations more efficient. We spent remarkably little time asking what makes them intelligent."
Artificial Intelligence has triggered one of the most significant technological shifts in modern business. And yet, the more I have read across management theory, cognitive science and the emerging practice of AI-enabled work over the past several months, the more I keep returning to a question that most AI conversations seem to bypass entirely.
Not: How can AI make organizations more intelligent?
But: What makes an organization intelligent in the first place?
The more I looked for an answer, the more startling the gap became. A century of management science has produced thousands of books explaining how to improve organizational performance. Far fewer attempt to explain what organizational intelligence actually is. Ask a room of experienced executives and you will get a dozen different answers — adaptive, data-driven, innovative, learning-oriented. All valuable properties. None of them is a definition. And that absence may be more consequential than it appears, because AI is now making it financially measurable.
The evidence hiding in plain sight
Toyota's factory floors and those of several of its closest competitors now run many of the same classes of AI tools — computer vision for quality inspection, predictive maintenance, sensor-driven anomaly detection. The technology gap between them is narrow. The results gap is not.
In 2022, Toyota's Production Digital Transformation Office built an internal AI platform on Google Cloud infrastructure, specifically designed to put machine learning directly into the hands of factory floor workers, regardless of technical background. By 2024, those workers had created 10,000 custom AI models, saving over 10,000 man-hours annually across Toyota's Japanese manufacturing network.¹
The platform itself wasn't exceptional. What was exceptional was the organization that received it: seven decades of Kaizen culture, the Jidoka principle of embedding human judgment into automated systems, a management philosophy that located problem-solving at the point of work rather than at the top of a hierarchy. When AI arrived, it found an organization already structured to absorb it.
Competitors who deployed comparable tools without that underlying architecture encountered what operational excellence practitioners have since described with unusual precision: they ended up digitising waste rather than eliminating it.² The AI was present. The organizational structure capable of acting on what it surfaced was not.
Toyota's VP of Digital Innovations, Jason Ballard, put the distinction plainly in a 2025 interview: "The differentiator isn't who has the best algorithm."³
This pattern extends well beyond manufacturing. McKinsey's Global Institute has documented substantial differences in the value organizations capture from AI deployments—differences that often exceed what improvements in model capability alone would predict. The limiting factor, it turns out, is rarely the technology.
A surprisingly old question
What's striking is that this question is not new — it's just been neglected.
In 1967, Harold Wilensky argued in Organizational Intelligence that organizations could be understood through their capacity to gather, process and use information, and identified the structural reasons most of them do it poorly. Decades later, Karl Weick, Ikujiro Nonaka, and others mapped how knowledge moves through organizations, how collective sensemaking shapes what an organization can even perceive, and why the same information produces different outcomes in different structures. More recently, Ethan Mollick has shown empirically that AI performance depends at least as much on the organizational context surrounding the technology as on the capabilities of the model itself.
These are fragments of a puzzle no one has fully assembled — perhaps because the stakes of leaving it unassembled once felt manageable. They no longer do.
The assumption nobody questioned
Management thinking has always located intelligence in individuals. Hire better talent. Develop better leaders. The AI conversation has simply shifted that frame: build better models, deploy more copilots, automate more decisions. Both approaches assume the same thing — that intelligence fundamentally belongs to individual agents, human or artificial.
But organizations are neither. They perceive through thousands of employees, customers, sensors, and systems simultaneously. They remember through documents, routines, software, and institutional knowledge that no single person fully holds. They learn through feedback loops that span years. They adapt — or don't — at rates no individual controls.
If intelligence emerges from the interaction of all these capabilities rather than from any single mind, then organizations may deserve to be studied as cognitive systems in their own right. Not as a metaphor. As an operational reality.
AI as amplifier
That reframing changes what AI actually is.
If organizations possess their own organizational cognitive architecture, AI doesn't give organizations intelligence—it amplifies whatever cognitive capabilities already exist within them. A high-perceiving organization, one with rich feedback loops and distributed decision-making, uses AI to perceive faster and more accurately. A low-perceiving one uses it to automate noise more efficiently. That's why AI returns don't follow a normal distribution. Organizations with stronger underlying cognitive architectures extract disproportionate value from the same technological increment.
The amplifier framing is not just descriptive. It's predictive. And if it holds, the most important AI decision a leadership team can make is not which model to choose — it's understanding the cognitive architecture that will use it.
A working hypothesis
Throughout this series I will explore a specific proposition: that organizations possess emergent cognitive capabilities — patterns of perceiving, remembering, learning, and coordinating — that function at the system level rather than the individual level, can be studied empirically, and can be deliberately designed and amplified. And that the interaction between these organizational capabilities and artificial intelligence produces something new — not smarter machines inside organizations, but a different kind of intelligence that belongs to neither humans nor machines alone.
I am calling this, provisionally, Organizational Co-Intelligence. The name may change. What won't is the intent: to test this rigorously, follow the evidence wherever it leads, and resist the temptation to dress a predetermined conclusion as research.
Good theories rarely begin with bold answers. They begin with better questions. If this series succeeds, it will not be because it introduces a new term. It will be because it helps us ask — with some precision and some evidence — a question that management has overlooked for decades.
What makes an organization truly intelligent?
That question deserves a serious investigation. This is where mine begins.
¹ Google Cloud Blog, How Toyota is revolutionizing manufacturing with AI
² Aicadium, Lean to AI: What Toyota's Transformation Looks Like Today (April 2026)
³ Jason Ballard, VP Digital Innovations, Toyota — Deloitte Insights (December 2025)
Explore the OCI series
Next: Why Organizations Need Intelligence
Discuss organizational co-intelligence in your organization.
Originally published on LinkedIn on July 6, 2026.
"We spent the last century learning how to make organizations more efficient. We spent remarkably little time asking what makes them intelligent."
Artificial Intelligence has triggered one of the most significant technological shifts in modern business. And yet, the more I have read across management theory, cognitive science and the emerging practice of AI-enabled work over the past several months, the more I keep returning to a question that most AI conversations seem to bypass entirely.
Not: How can AI make organizations more intelligent?
But: What makes an organization intelligent in the first place?
The more I looked for an answer, the more startling the gap became. A century of management science has produced thousands of books explaining how to improve organizational performance. Far fewer attempt to explain what organizational intelligence actually is. Ask a room of experienced executives and you will get a dozen different answers — adaptive, data-driven, innovative, learning-oriented. All valuable properties. None of them is a definition. And that absence may be more consequential than it appears, because AI is now making it financially measurable.
The evidence hiding in plain sight
Toyota's factory floors and those of several of its closest competitors now run many of the same classes of AI tools — computer vision for quality inspection, predictive maintenance, sensor-driven anomaly detection. The technology gap between them is narrow. The results gap is not.
In 2022, Toyota's Production Digital Transformation Office built an internal AI platform on Google Cloud infrastructure, specifically designed to put machine learning directly into the hands of factory floor workers, regardless of technical background. By 2024, those workers had created 10,000 custom AI models, saving over 10,000 man-hours annually across Toyota's Japanese manufacturing network.¹
The platform itself wasn't exceptional. What was exceptional was the organization that received it: seven decades of Kaizen culture, the Jidoka principle of embedding human judgment into automated systems, a management philosophy that located problem-solving at the point of work rather than at the top of a hierarchy. When AI arrived, it found an organization already structured to absorb it.
Competitors who deployed comparable tools without that underlying architecture encountered what operational excellence practitioners have since described with unusual precision: they ended up digitising waste rather than eliminating it.² The AI was present. The organizational structure capable of acting on what it surfaced was not.
Toyota's VP of Digital Innovations, Jason Ballard, put the distinction plainly in a 2025 interview: "The differentiator isn't who has the best algorithm."³
This pattern extends well beyond manufacturing. McKinsey's Global Institute has documented substantial differences in the value organizations capture from AI deployments—differences that often exceed what improvements in model capability alone would predict. The limiting factor, it turns out, is rarely the technology.
A surprisingly old question
What's striking is that this question is not new — it's just been neglected.
In 1967, Harold Wilensky argued in Organizational Intelligence that organizations could be understood through their capacity to gather, process and use information, and identified the structural reasons most of them do it poorly. Decades later, Karl Weick, Ikujiro Nonaka, and others mapped how knowledge moves through organizations, how collective sensemaking shapes what an organization can even perceive, and why the same information produces different outcomes in different structures. More recently, Ethan Mollick has shown empirically that AI performance depends at least as much on the organizational context surrounding the technology as on the capabilities of the model itself.
These are fragments of a puzzle no one has fully assembled — perhaps because the stakes of leaving it unassembled once felt manageable. They no longer do.
The assumption nobody questioned
Management thinking has always located intelligence in individuals. Hire better talent. Develop better leaders. The AI conversation has simply shifted that frame: build better models, deploy more copilots, automate more decisions. Both approaches assume the same thing — that intelligence fundamentally belongs to individual agents, human or artificial.
But organizations are neither. They perceive through thousands of employees, customers, sensors, and systems simultaneously. They remember through documents, routines, software, and institutional knowledge that no single person fully holds. They learn through feedback loops that span years. They adapt — or don't — at rates no individual controls.
If intelligence emerges from the interaction of all these capabilities rather than from any single mind, then organizations may deserve to be studied as cognitive systems in their own right. Not as a metaphor. As an operational reality.
AI as amplifier
That reframing changes what AI actually is.
If organizations possess their own organizational cognitive architecture, AI doesn't give organizations intelligence—it amplifies whatever cognitive capabilities already exist within them. A high-perceiving organization, one with rich feedback loops and distributed decision-making, uses AI to perceive faster and more accurately. A low-perceiving one uses it to automate noise more efficiently. That's why AI returns don't follow a normal distribution. Organizations with stronger underlying cognitive architectures extract disproportionate value from the same technological increment.
The amplifier framing is not just descriptive. It's predictive. And if it holds, the most important AI decision a leadership team can make is not which model to choose — it's understanding the cognitive architecture that will use it.
A working hypothesis
Throughout this series I will explore a specific proposition: that organizations possess emergent cognitive capabilities — patterns of perceiving, remembering, learning, and coordinating — that function at the system level rather than the individual level, can be studied empirically, and can be deliberately designed and amplified. And that the interaction between these organizational capabilities and artificial intelligence produces something new — not smarter machines inside organizations, but a different kind of intelligence that belongs to neither humans nor machines alone.
I am calling this, provisionally, Organizational Co-Intelligence. The name may change. What won't is the intent: to test this rigorously, follow the evidence wherever it leads, and resist the temptation to dress a predetermined conclusion as research.
Good theories rarely begin with bold answers. They begin with better questions. If this series succeeds, it will not be because it introduces a new term. It will be because it helps us ask — with some precision and some evidence — a question that management has overlooked for decades.
What makes an organization truly intelligent?
That question deserves a serious investigation. This is where mine begins.
¹ Google Cloud Blog, How Toyota is revolutionizing manufacturing with AI
² Aicadium, Lean to AI: What Toyota's Transformation Looks Like Today (April 2026)
³ Jason Ballard, VP Digital Innovations, Toyota — Deloitte Insights (December 2025)
Explore the OCI series
Next: Why Organizations Need Intelligence
Discuss organizational co-intelligence in your organization.