AI is making individuals more productive, but many organisations are not becoming more effective. This article explores why. Drawing on research and 16 years of practical experience, it argues that the biggest barrier to performance is often not people or technology, but the system in which they work. The solution is not more output. It is better work design.
AI is exposing a problem that technology cannot solve: many organisations have optimised the components of work while neglecting the human and organisational system that must connect them.
For decades, organisations have invested heavily in physical capital, financial capital and, more recently, digital capital.
Now attention is turning to another productive asset: the human brain.
The McKinsey Health Institute defines “brain capital” as the combination of brain health and brain skills [1].
These skills include the cognitive, interpersonal and self-leadership capabilities that enable people to adapt, relate and contribute meaningfully. Its argument is compelling: as artificial intelligence becomes more capable, human judgement, empathy, resilience, creativity and purpose become more important, not less.
But there is an uncomfortable contradiction at the heart of this idea.
Many organisations say that people are their greatest asset while structuring work in ways that steadily deplete their attention, energy, clarity and capacity to think.
This is not principally a resilience problem.
It is a work-design problem.
AI has made the contradiction visible.
AI can make an individual task faster without making the organisation more effective. A survey of 6,000 digital workers, reported in the Inc. article “AI Is Making Workers More Productive. So Why Aren’t Companies Performing Better?”, found that 75 per cent believed AI had made them more productive and that it automated an average of 11 hours of work a week [2]. Yet only 13 per cent said it had meaningfully improved their organisation’s overall performance [2].
Workers also reported spending 6.4 hours a week providing context to AI, supervising its output and correcting problems downstream (botsitting) [2].
The issue is not necessarily that AI fails to save time. It is that the surrounding system often fails to convert saved time into organisational value.
Faster production can result in:
Local efficiency can therefore increase systemic inefficiency.
The Inc. article describes this as coordination neglect, a term developed by organisational scholars Chip Heath and Nancy Staudenmayer [3]. It refers to our tendency to underestimate the effort required to integrate work after it has been divided into roles, teams and specialist components.
This matters because the value of most knowledge work does not reside in isolated tasks. It emerges from the relationships between them.
A marketing team can produce a campaign faster. That creates no value if Legal cannot review it, Sales cannot use it or customers do not find it relevant.
A product team can generate more ideas. That is not progress if leadership cannot make timely choices about which ideas deserve scarce engineering capacity.
An employee can produce twice as much. That may make the wider organisation slower if everyone downstream must absorb, interpret, check and coordinate considerably more material.
The unit of performance is therefore not the person or task. It is the end-to-end flow of value.
AI has not created coordination neglect. It has amplified it.
Modern organisations did not become fragmented by accident.
Management has long relied on decomposition. Break the strategy into objectives. Divide the objectives into functions. Divide the functions into departments. Divide the departments into roles. Give each component a budget, target and accountable leader.
This logic has delivered enormous benefits. Specialisation creates expertise. Clear divisions can make large organisations manageable. Financial discipline matters. Matrix structures can enable organisations to share scarce expertise and connect functions with markets, products and geographies.
The problem begins when decomposition is treated as the whole act of management. Dividing work is only half the design challenge. The other half is integrating it.
Business schools did not invent organisational dysfunction. But influential traditions within management education may have reinforced the intellectual habits that sustain it. MBA curricula have been criticised for dividing management into functional disciplines without adequately teaching students how those disciplines should be reintegrated [4].
Business schools have also been accused of privileging abstract financial, economic and statistical models over the complex and often unquantifiable realities of leading human organisations [5].
The result may be leaders who are highly capable of optimising the pieces, yet less practised in designing the relationships between them.
Influential critics have argued that agency theory, neoclassical economics and shareholder-value maximisation gained disproportionate influence within management education and helped legitimise particular approaches to governance, incentives and control [6].
That distinction matters.
An organisation can contain excellent functions and still perform poorly as a system.
Every department can meet its targets while the customer waits, decisions stall, work is duplicated and employees spend their days coordinating around the structure.
The literature on systems thinking supports this distinction between analysing individual parts and understanding the relationships, feedback and underlying structures that shape the performance of the whole.
These decisions are often rational within each component. Together, they can become organisationally irrational.
This criticism of management thinking is not new.
William Edwards Deming argued that organisations should be understood as systems of interdependent components [7]. Management’s responsibility is not to maximise each component independently, but to improve the system through which the components create value together [7].
Deming estimated from his experience that approximately 94 per cent of problems and opportunities for improvement belonged to the system, and were therefore management’s responsibility, while approximately 6 per cent arose from special causes [7].
This was an estimate rather than a universal statistical law, but the implication remains powerful: when capable people repeatedly struggle, leaders should investigate the conditions in which they work before attributing the problem to individual effort or capability.
Deming also identified what he called the Seven Deadly Diseases of Management. These included:
These warnings remain strikingly relevant. Organisations continue to divide work into functions, give each function its own targets and assess people through visible measures of individual activity. Each component may appear productive while the organisation as a whole becomes slower, more expensive and more exhausting.
AI can compound this error. It produces highly visible measures of local productivity: documents generated, tasks completed, code produced and hours apparently saved. Far less visible are the additional checking, coordination, queues, rework and cognitive load created elsewhere.
We optimise the component we can measure while overlooking the system that creates the value.
This is not an argument against MBAs, analytical rigour or financial discipline. It is an argument against mistaking decomposition for management. Dividing work into components is only half the task. The other half is creating the clarity, capacity and coordination through which the components produce value together.
The more credible criticism is therefore one of imbalance.
Conventional management education can develop leaders who are highly capable of analysing, allocating and optimising separate components without cultivating an equally strong understanding of relationships, variation, psychology and system-level consequences.
Dividing work is not the same as managing it.
Measurement is not the same as understanding it.
And improving every part does not necessarily improve the whole.
Matrix structures were intended to overcome silos. In practice, a poorly designed matrix can connect every silo to every other silo without removing any of them.
The result is not integration. It is coordination overload.
McKinsey & Company and Gallup research has found both strengths and weaknesses in matrix organisations [7]. Matrixed employees reported benefits in collaboration, recognition and innovation. They were also less clear about expectations and more likely to spend their time responding to colleagues and attending internal meetings. The research identifies role clarity and accountability as particularly important drivers of organisational health [8].
A later Gallup analysis found that one-third of highly matrixed employees said they spent most of their day in internal meetings, while 45 per cent said they spent most of it responding to requests from colleagues [9]. It identifies three recurring problems: cognitive overload, role conflict and ambiguity, and coordination failure [9].
This creates a predictable cycle:
The organisation responds with more alignment, reporting and control. The intervention becomes another source of the problem.
Research on organisational silos reinforces the broader point. A scoping review found that silos can become barriers to organisational goals by threatening internal cooperation [10].
The implication is not that every matrix should be dismantled. It is that an organisational chart cannot solve the coordination problem. The lived operating system must be deliberately designed.
Brain capital is not merely something people possess.
This is where the brain-capital conversation needs to go further.
Brain capital is often discussed as an asset within the individual: cognitive capability, emotional regulation, resilience, adaptability and interpersonal skill.
These capabilities matter. But their expression is highly dependent on conditions.
A skilled brain in a poorly designed system cannot indefinitely compensate for:
The organisation does not simply employ brain capital. It either multiplies it or depletes it.
This is why wellbeing and performance should not be managed as separate agendas. They share many of the same operating conditions.
In my work, I bring these together as five levers:
Are priorities, roles, outcomes and expectations sufficiently clear for people to make good decisions without repeatedly seeking permission or alignment?
Do people have meaningful autonomy over their workload, calendar, attention and method of working?
Do relationships support trust, belonging, challenge, coordination and honest conversations about capacity?
Are attention, movement, sleep and recovery treated as foundations of performance, or as private responsibilities to be managed outside working hours?
Do leadership behaviour, incentives and everyday norms reinforce healthy effectiveness, or do they reward urgency, availability and visible busyness?
Most teams do not have an effort problem. They have a work systems problem.
Your organisation cannot claim to invest in brain capital while quietly rewarding behaviours that deplete it.
A healthier and more effective system begins by separating three concepts that are often confused.
Organisational performance depends on completing the right combinations of activities, across boundaries, without creating more friction, risk or human cost than the value produced.
This changes the questions leaders should ask.
Instead of:
Ask:
The Inc. analysis reports that employees in organisations measuring both productivity and quality were more likely to say that AI had improved their work quality than those in organisations measuring productivity alone [2]. Its broader recommendation is to measure the entire workflow, rather than the isolated task touched by AI (no surprise there).
That is an important distinction.
We need a healthier and more effective organisational operating system.
The answer is not another wellbeing benefit added to an overloaded system.
It is to redesign how work happens.
The approach I have developed and proven over 16 years, begins with a simple weekly rhythm: Prioritise, Learn and Plan.
When practised by a team, this reflection becomes a decentralised system of organisational design.
The team examines whether calendars reflect stated priorities, whether coordination is producing value, whether workloads are realistic, whether meetings are necessary, and whether people have the capacity to perform the work expected of them.
A healthy and effective way of working provides the practical structure for prioritising, reviewing value, removing non-value-adding work and planning focus, meeting, responsive and recharge time.
Over time, this rhythm can establish healthier and more effective norms:
This supports both individual capacity and organisational coordination.
It also provides a better foundation for AI. Once an organisation understands the workflow, dependencies, constraints and desired outcomes, it can automate intelligently. Without that understanding, it may simply automate the creation of more work.
The brain-capital agenda is important because it reframes human health and capability as central to economic and organisational strength.
But an organisation cannot yoga, coach or app its way out of conflicting priorities, meeting overload and poor work design.
Nor can it automate its way out with AI.
The strategic opportunity is to redesign work so that technology, structure and human capability reinforce one another.
That means treating:
The organisations that succeed in the era of AI, will be those that become better at deciding what is worth producing, coordinating the work required to produce it and protecting the human capacity on which sound judgement, relationships and innovation depend.
That is how brain capital becomes more than an attractive concept.
It becomes an operating principle.
This may require leaders to confront a difficult truth: The greatest barrier to human performance is sometimes not the capability of the people or the technology available to them.
It is the system in which they are being asked to work.
[1] https://www.linkedin.com/pulse/brain-capital-investing-human-mckinseyhealthinstitute-2ftoc/
[3] https://www.sciencedirect.com/science/article/abs/pii/S0191308500220054
[4] https://gbcroundtable.org/files/Navarro%202008%281%29.pdf
[5] https://ceo.usc.edu/wp-content/uploads/2005/02/2005_06-g05_6-Why_Business_Schools_Lost_Way.pdf
[6] https://www.jstor.org/stable/40214265
[7] Deming, W. E. (2018). The new economics for industry, government, education. MIT press.
[9] https://www.gallup.com/workplace/354935/teams-bosses-overcoming-matrix-madness.aspx
[10] https://www.mdpi.com/2075-4698/10/3/56