Originally published on LinkedIn on August 18, 2026.
Why Organizations That Communicate Still Fail to Think Together
"The limits of my language mean the limits of my world."
— Ludwig Wittgenstein, Tractatus Logico-Philosophicus
The previous article ended with a proposition that deserves a direct answer.
If organizational intelligence doesn't reside in technology, and if it doesn't belong to any individual, then where exactly does it live? What is the invisible substrate through which thousands of specialized people develop the capacity to think together?
The answer, I believe, has been hiding in plain sight. Not in systems. Not in hierarchy. Not even primarily in people. It lives in something those people share — something so embedded in the daily life of organizations that most never notice it until it's gone. It lives in shared meaning.
The hidden consequence of specialization
The previous articles established that specialization creates an irreversible need for integration. As organizations grow, knowledge becomes distributed. Engineering understands one slice of reality. Sales understands another. Finance sees something different. Operations sees something different still.
This is not a flaw — it is the source of organizational capability. But specialization creates a consequence that most organizations never fully reckon with.
When knowledge fragments, so does reality. The moment specialization deepens, no one in the organization experiences it exactly as it is. Everyone holds a partial version, filtered through the concepts, priorities, and language of their function. And when people carry different versions of reality into the same room, they can exchange information fluently while understanding almost nothing in common.
Living systems have confronted this problem for far longer than organizations have. The immune system — perhaps the most striking example of distributed intelligence in biology — coordinates across billions of highly specialized cells without any central command. Individual T-cells and B-cells possess narrow, dedicated capabilities. They coordinate not through hierarchy but through shared molecular representations: the chemical signals that allow distributed cells to recognize a threat and respond coherently across an entire organism. When that shared signaling breaks down, the system does not simply become less effective. It begins attacking the organism it was built to protect. The biological lesson is precise: in any complex system built on specialization, distributed intelligence requires shared representations. Without them, specialization produces pathology rather than capability.
The gap is not between what people know. It is between what the same words mean to different people.
What the meeting room actually reveals
Consider a leadership team discussing customer experience.
The phrase appears straightforward. Everyone around the table understands it — or believes they do. The product leader thinks about usability. Marketing thinks about brand perception. Operations thinks about service consistency. Finance thinks about retention and lifetime value. The words are identical. The meanings are not.
Communication occurred. Information was exchanged. The meeting was productive by most organizational standards. And yet each participant left carrying a different interpretation of what was actually decided.
This happens across every concept that matters at scale: innovation, quality, alignment, transformation, strategy, agility. The larger and more specialized an organization becomes, the more likely these words are to fragment into local meanings. And when meaning fragments, coordination quietly follows.
Karl Weick's decades of research on organizational sensemaking established the distinction precisely: organizations do not merely process information.5 They continuously construct meaning. People act not on reality itself, but on their interpretation of it. Coordination therefore depends less on information transfer than on the creation of sufficiently shared understanding — a very different problem from the ones most organizations have been trying to solve.
The operating system beneath the organization
This leads to what I believe is the most consequential idea in this series so far.
Organizations do not coordinate through information. They coordinate through shared meaning.
At first glance, the distinction appears semantic. It is not. Two teams can possess identical information and make incompatible decisions. Two executives can agree on the same strategy and pursue conflicting priorities. Two departments can review the same dashboard and reach opposite conclusions. The missing variable is not information. It is interpretation.
Shared meaning functions as a kind of organizational operating system. It allows distributed knowledge to become collectively intelligible. Without it, specialization produces fragmentation. With it, specialization produces intelligence.
Every organization develops, over time, a set of concepts that determine how reality is interpreted. What is a customer? What constitutes quality? What does innovation actually mean here? These concepts become embedded in language, principles, frameworks, metrics, and institutional stories. Eventually they form something larger — an invisible architecture of meaning that shapes almost every decision the organization makes, usually without anyone recognizing it as architecture at all. Mission statements, operating principles, governance models, and shared taxonomies are doing something deeper than administration. They are attempts to make understanding transferable. What many organizations dismiss as bureaucracy is often a primitive and undervalued form of cognitive infrastructure.
Organizations spend billions designing technological, financial, and operational infrastructure. Yet the infrastructure that determines whether specialized knowledge becomes collective intelligence is rarely designed at all.
What AI exposed that experience had concealed
For decades, organizations compensated for fragmented meaning through people — specifically, through experienced managers and informal translators who built bridges between specialized realities through context, relationships, and accumulated judgment. AI has made the cost of that dependency visible.
In June 2026, Mark Zuckerberg acknowledged in an internal memo that Meta had made significant mistakes in its AI-driven workforce transformation.1 The scale was considerable: roughly 8,000 employees laid off, another 7,000 transferred into AI-focused roles, affecting close to 20% of its workforce. "Given the complexity of these changes," Zuckerberg wrote, "we've made mistakes and will almost certainly make more."
Most observers interpreted this as a story about the difficulty of AI transformation. I believe it reveals something more specific.
What Meta and many organizations like it discovered was that the people they cut were not simply performing tasks that AI could replicate. They were carrying something AI could not: the interpretive infrastructure through which the organization made sense of itself. The institutional context. The informal translation between functions. The accumulated judgment about what things actually mean in this organization, in this market, at this moment.
The evidence is accumulating at scale. A Robert Half survey conducted in April 2026 found that 32% of U.S. hiring managers who eliminated a role primarily due to AI later rehired for the same or similar position.2 Orgvue's annual workforce study found that 55% of business leaders who reduced headcount during an AI rollout subsequently admitted those decisions had been wrong.3 Ford rehired hundreds of experienced engineers after automated systems failed to handle quality issues that required the contextual judgment those engineers carried.4 In each case, the reversal followed the same discovery: the capability that had been cut was not task performance. It was organizational memory.
What organizations lost was not labor capacity. It was meaning-making capacity. And as those reversals demonstrate, meaning-making capacity cannot be rebuilt on demand.
A different view of leadership
This perspective changes how we think about leadership as well.
If organizations coordinate through shared meaning, then one of leadership's most consequential responsibilities is the construction and maintenance of shared reality — the clarification of concepts, the cultivation of common language, the alignment of interpretation across an increasingly specialized system.
This is not about telling people what to think. It is about creating the conditions from which people can think compatibly.
Organizations that appear highly aligned may not communicate more than others. They may simply spend less energy translating between competing realities. Their concepts are clearer. Their assumptions are more visible. Their meaning travels further with less friction. What appears from the outside as culture, execution, or leadership is often something more foundational — a shared reality that allows collective intelligence to emerge.
What AI is actually amplifying
Artificial intelligence did not create the shared meaning problem. It exposed it.
Language models operate almost entirely through representations — concepts, language, definitions, context. As AI becomes embedded in organizational work, it encounters the organization's shared meaning directly. It learns from it, amplifies it, and increasingly acts through it. An organization with fragmented concepts tends to generate fragmented outputs. An organization with coherent shared understanding tends to generate more coherent ones. The technology is often identical. The organizational reality being amplified is not.
The question facing every leadership team is therefore not only which AI to deploy. It is what kind of shared reality their organization has actually built — and whether that reality is one worth amplifying.
How shared meaning is constructed, why it erodes under pressure, and what it takes to build it deliberately are where this investigation turns next. But a question waits just beyond that one. If organizational co-intelligence is real and not merely metaphorical, it should be observable — the way intelligence in living systems is observable through behavior: how a system responds to novelty, recovers from disruption, acts coherently under pressure. Organizations leave the same kind of traces. We have simply not known to look for them.
───
Selected References
Enterprise Evidence
1. Reuters. Zuckerberg Says Meta Made "Mistakes" in AI Workforce Shift (June 12, 2026).
2. Robert Half. AI Workforce Impact Survey, April 2026. Cited in CNBC, "Employers Who Laid Off Workers Citing AI Are Already Reversing Their Decisions" (July 1, 2026).
3. Orgvue. Annual Workforce Study (2026). Cited in multiple sources including Forrester Predictions 2026.
4. CNBC. Employers Who Laid Off Workers Citing AI Are Already Reversing Their Decisions (July 1, 2026).
Foundations of Organizational Thought
5. Weick, K. E. (1995). Sensemaking in Organizations.
6. Weick, K. E., Sutcliffe, K. M., & Obstfeld, D. (2005). Organizing and the Process of Sensemaking.
7. Nonaka, I., & Takeuchi, H. (1995). The Knowledge-Creating Company.
8. Maitlis, S., & Christianson, M. (2014). Sensemaking in Organizations: Taking Stock and Moving Forward.
9. Simon, H. A. (1947). Administrative Behavior.
10. March, J. G., & Simon, H. A. (1958). Organizations.
11. Hayek, F. A. (1945). The Use of Knowledge in Society.
Explore the OCI series
Next: What Intelligence Leaves Behind
Discuss organizational co-intelligence in your organization.
Originally published on LinkedIn on August 18, 2026.
Why Organizations That Communicate Still Fail to Think Together
"The limits of my language mean the limits of my world."
— Ludwig Wittgenstein, Tractatus Logico-Philosophicus
The previous article ended with a proposition that deserves a direct answer.
If organizational intelligence doesn't reside in technology, and if it doesn't belong to any individual, then where exactly does it live? What is the invisible substrate through which thousands of specialized people develop the capacity to think together?
The answer, I believe, has been hiding in plain sight. Not in systems. Not in hierarchy. Not even primarily in people. It lives in something those people share — something so embedded in the daily life of organizations that most never notice it until it's gone. It lives in shared meaning.
The hidden consequence of specialization
The previous articles established that specialization creates an irreversible need for integration. As organizations grow, knowledge becomes distributed. Engineering understands one slice of reality. Sales understands another. Finance sees something different. Operations sees something different still.
This is not a flaw — it is the source of organizational capability. But specialization creates a consequence that most organizations never fully reckon with.
When knowledge fragments, so does reality. The moment specialization deepens, no one in the organization experiences it exactly as it is. Everyone holds a partial version, filtered through the concepts, priorities, and language of their function. And when people carry different versions of reality into the same room, they can exchange information fluently while understanding almost nothing in common.
Living systems have confronted this problem for far longer than organizations have. The immune system — perhaps the most striking example of distributed intelligence in biology — coordinates across billions of highly specialized cells without any central command. Individual T-cells and B-cells possess narrow, dedicated capabilities. They coordinate not through hierarchy but through shared molecular representations: the chemical signals that allow distributed cells to recognize a threat and respond coherently across an entire organism. When that shared signaling breaks down, the system does not simply become less effective. It begins attacking the organism it was built to protect. The biological lesson is precise: in any complex system built on specialization, distributed intelligence requires shared representations. Without them, specialization produces pathology rather than capability.
The gap is not between what people know. It is between what the same words mean to different people.
What the meeting room actually reveals
Consider a leadership team discussing customer experience.
The phrase appears straightforward. Everyone around the table understands it — or believes they do. The product leader thinks about usability. Marketing thinks about brand perception. Operations thinks about service consistency. Finance thinks about retention and lifetime value. The words are identical. The meanings are not.
Communication occurred. Information was exchanged. The meeting was productive by most organizational standards. And yet each participant left carrying a different interpretation of what was actually decided.
This happens across every concept that matters at scale: innovation, quality, alignment, transformation, strategy, agility. The larger and more specialized an organization becomes, the more likely these words are to fragment into local meanings. And when meaning fragments, coordination quietly follows.
Karl Weick's decades of research on organizational sensemaking established the distinction precisely: organizations do not merely process information.5 They continuously construct meaning. People act not on reality itself, but on their interpretation of it. Coordination therefore depends less on information transfer than on the creation of sufficiently shared understanding — a very different problem from the ones most organizations have been trying to solve.
The operating system beneath the organization
This leads to what I believe is the most consequential idea in this series so far.
Organizations do not coordinate through information. They coordinate through shared meaning.
At first glance, the distinction appears semantic. It is not. Two teams can possess identical information and make incompatible decisions. Two executives can agree on the same strategy and pursue conflicting priorities. Two departments can review the same dashboard and reach opposite conclusions. The missing variable is not information. It is interpretation.
Shared meaning functions as a kind of organizational operating system. It allows distributed knowledge to become collectively intelligible. Without it, specialization produces fragmentation. With it, specialization produces intelligence.
Every organization develops, over time, a set of concepts that determine how reality is interpreted. What is a customer? What constitutes quality? What does innovation actually mean here? These concepts become embedded in language, principles, frameworks, metrics, and institutional stories. Eventually they form something larger — an invisible architecture of meaning that shapes almost every decision the organization makes, usually without anyone recognizing it as architecture at all. Mission statements, operating principles, governance models, and shared taxonomies are doing something deeper than administration. They are attempts to make understanding transferable. What many organizations dismiss as bureaucracy is often a primitive and undervalued form of cognitive infrastructure.
Organizations spend billions designing technological, financial, and operational infrastructure. Yet the infrastructure that determines whether specialized knowledge becomes collective intelligence is rarely designed at all.
What AI exposed that experience had concealed
For decades, organizations compensated for fragmented meaning through people — specifically, through experienced managers and informal translators who built bridges between specialized realities through context, relationships, and accumulated judgment. AI has made the cost of that dependency visible.
In June 2026, Mark Zuckerberg acknowledged in an internal memo that Meta had made significant mistakes in its AI-driven workforce transformation.1 The scale was considerable: roughly 8,000 employees laid off, another 7,000 transferred into AI-focused roles, affecting close to 20% of its workforce. "Given the complexity of these changes," Zuckerberg wrote, "we've made mistakes and will almost certainly make more."
Most observers interpreted this as a story about the difficulty of AI transformation. I believe it reveals something more specific.
What Meta and many organizations like it discovered was that the people they cut were not simply performing tasks that AI could replicate. They were carrying something AI could not: the interpretive infrastructure through which the organization made sense of itself. The institutional context. The informal translation between functions. The accumulated judgment about what things actually mean in this organization, in this market, at this moment.
The evidence is accumulating at scale. A Robert Half survey conducted in April 2026 found that 32% of U.S. hiring managers who eliminated a role primarily due to AI later rehired for the same or similar position.2 Orgvue's annual workforce study found that 55% of business leaders who reduced headcount during an AI rollout subsequently admitted those decisions had been wrong.3 Ford rehired hundreds of experienced engineers after automated systems failed to handle quality issues that required the contextual judgment those engineers carried.4 In each case, the reversal followed the same discovery: the capability that had been cut was not task performance. It was organizational memory.
What organizations lost was not labor capacity. It was meaning-making capacity. And as those reversals demonstrate, meaning-making capacity cannot be rebuilt on demand.
A different view of leadership
This perspective changes how we think about leadership as well.
If organizations coordinate through shared meaning, then one of leadership's most consequential responsibilities is the construction and maintenance of shared reality — the clarification of concepts, the cultivation of common language, the alignment of interpretation across an increasingly specialized system.
This is not about telling people what to think. It is about creating the conditions from which people can think compatibly.
Organizations that appear highly aligned may not communicate more than others. They may simply spend less energy translating between competing realities. Their concepts are clearer. Their assumptions are more visible. Their meaning travels further with less friction. What appears from the outside as culture, execution, or leadership is often something more foundational — a shared reality that allows collective intelligence to emerge.
What AI is actually amplifying
Artificial intelligence did not create the shared meaning problem. It exposed it.
Language models operate almost entirely through representations — concepts, language, definitions, context. As AI becomes embedded in organizational work, it encounters the organization's shared meaning directly. It learns from it, amplifies it, and increasingly acts through it. An organization with fragmented concepts tends to generate fragmented outputs. An organization with coherent shared understanding tends to generate more coherent ones. The technology is often identical. The organizational reality being amplified is not.
The question facing every leadership team is therefore not only which AI to deploy. It is what kind of shared reality their organization has actually built — and whether that reality is one worth amplifying.
How shared meaning is constructed, why it erodes under pressure, and what it takes to build it deliberately are where this investigation turns next. But a question waits just beyond that one. If organizational co-intelligence is real and not merely metaphorical, it should be observable — the way intelligence in living systems is observable through behavior: how a system responds to novelty, recovers from disruption, acts coherently under pressure. Organizations leave the same kind of traces. We have simply not known to look for them.
───
Selected References
Enterprise Evidence
1. Reuters. Zuckerberg Says Meta Made "Mistakes" in AI Workforce Shift (June 12, 2026).
2. Robert Half. AI Workforce Impact Survey, April 2026. Cited in CNBC, "Employers Who Laid Off Workers Citing AI Are Already Reversing Their Decisions" (July 1, 2026).
3. Orgvue. Annual Workforce Study (2026). Cited in multiple sources including Forrester Predictions 2026.
4. CNBC. Employers Who Laid Off Workers Citing AI Are Already Reversing Their Decisions (July 1, 2026).
Foundations of Organizational Thought
5. Weick, K. E. (1995). Sensemaking in Organizations.
6. Weick, K. E., Sutcliffe, K. M., & Obstfeld, D. (2005). Organizing and the Process of Sensemaking.
7. Nonaka, I., & Takeuchi, H. (1995). The Knowledge-Creating Company.
8. Maitlis, S., & Christianson, M. (2014). Sensemaking in Organizations: Taking Stock and Moving Forward.
9. Simon, H. A. (1947). Administrative Behavior.
10. March, J. G., & Simon, H. A. (1958). Organizations.
11. Hayek, F. A. (1945). The Use of Knowledge in Society.
Explore the OCI series
Next: What Intelligence Leaves Behind
Discuss organizational co-intelligence in your organization.