Sangeet Paul Choudary is the author of “Platform Revolution”, a book I wasn’t that impressed with, as described in my review here. Gladly, I do not feel the same way about “Reshuffle: Who Wins When AI Restacks the Knowledge Economy.” It is a difficult book to read, but I would say it is nearly essential if you are in the world of AI. The author’s point is that, with AI, it is not about the roadmap; it is about the terrain (or map) itself: designing the new game and its rules.
Solar
For example, in the 2010s, residential solar technology had become efficient and affordable. The breakthrough came not through better panels, but through better financing. Adoption was unlocked by solving the system’s constraints, not by improving the tool’s performance.
It is not abut intelligence
The idea that AI will suddenly leap into superintelligence distracts us from the real work: evolving our systems to keep pace with today’s capabilities. Crucially, AI is not about intelligence, reasoning, or meaning. Instead, it involves processing vast amounts of data and identifying statistical patterns. Even tasks that seem to require comprehension—like interpreting language, recognising images, or generating realistic pictures—are performed using this same underlying mechanism.
Five functions
More specifically, AI tools perform the same five core functions: they sense the environment, build a working model of the world around them, reason and act based on that model, and continually learn and update it.
Recognising the patterns
AI tools may not understand meaning as humans do, but they learn by recognising patterns and familiar language structures. Unlike older software, modern AI can selectively attend to what matters in the moment, guided by memory.
Restructuring the system
As a result, the real impact of AI comes not from how it performs a task, but from how it restructures the entire system around that task. When AI enters a system, it alters the economic logic. It changes how value is created. In short, it’s not the technology, it’s the system.
Everything explicit
In this context, much of an organisation’s most valuable knowledge is tacit, buried in meetings, emails, and conversations—making it inaccessible and unshared. AI helps convert such tacit knowledge into explicit, actionable insights that can be utilised across an organisation.
Finding the right information at the right time
If you’ve ever worked in a large organisation, you know the real challenge isn’t finding information—it’s finding the right information at the right time. This highlights the crucial role of organisational knowledge management. Organisational knowledge—the collective understanding and expertise across teams—enables collaboration, even when teams focus on different things. Without knowledge sharing, teams end up working in isolation, duplicating efforts and slowing the entire organisation down.
Manage knowledge and allocate attention
This is why AI is becoming an institutional infrastructure that shapes how organisations manage knowledge and allocate attention, and ultimately, how they make decisions. The more relevant question, then, is not whether AI can think like us, but how it changes how we think, decide, and coordinate.
It is the terrain, not the map
Consequently, competitive edge no longer comes from owning or adopting the best tool, but from understanding how that tool restructures the entire playing field.
AI alters the system
To understand the true impact of any new technology on an economy, an organisation, or society at large, we must first examine how it alters the fundamental interactions within the system. The book uses COVID, Singapore, containers, AirBnB, Stripe, Tesla, Shein, Google Maps, talents spotting in the NFL, kitchens, KIVA, music, ZARA, barcodes, Walmart, the British Empire, agriculture, Uber, Tuareg, magicians, Jeff Bezos, Real Madrid, Barcelona, NASA, Shopfiy, NotebookLM, Palantir, MrBeast, Muji, vibe coding, application wrappers, Formula 1, TikTok, Best Buy, Sephora and many more. You should study to understand the points the author is making.
Coordination eating complexity
It isn’t software that is eating the world – it is coordination eating complexity. Because coordination is the central driver of value in today’s economy, AI’s ability to model a domain and align intent and action reliably with that model makes it particularly well-suited for coordination. What makes AI truly transformative is not its performance on benchmarks of intelligence but rather its ability to model, mediate, and move fragmented systems toward alignment.
The coordination gap – and why AI matters
The coordination gap is a divide between what today’s coordination mechanisms manage well and what most economic activity actually requires. This is precisely why artificial intelligence (AI) matters – not so much as a tool for improving productivity through automation, but as a mechanism for coordinating what has so far remained uncoordinated and fragmented. AI observes the world, creates a working model of it, reasons through possible choices based on that model, acts on those decisions, and learns from the outcomes to improve over time. It helps bridge the gap between well-coordinated, structured systems and the vast, unstructured processes and workflows that rely on tacit knowledge.
It is the system, stupid
AI’s real power lies not in automating individual tasks but in coordinating entire systems and reconfiguring how they operate. AI reduces the cost of task execution. Equally important, but far less discussed, is AI’s potential to lower coordination costs: AI can help bridge this gap by making sense of the unstructured information each player holds, building a shared model across them, and delivering the right insights to the right people at the right time. In a multi-team sales process, AI can consolidate conversations. When a supplier changes delivery timelines, AI can detect the change in one system and automatically update forecasts, procurement schedules, and downstream planning across partner companies, avoiding a cascade of miscommunication.
It’s about coordination
Firms that focus on automation may achieve short-term gains, but those using AI to orchestrate complex systems will create new forms of value and gain a competitive edge. AI’s primary strength lies in coordination. Its ability to coordinate without requiring upfront consensus, combined with improved decision-making, execution, governance, learning, and adaptation, forms a self-reinforcing cycle. Enhanced coordination among components leads to better performance, and greater specialisation then increases the need for further coordination.
The coordination paradox
For all its promise, AI can also make coordination worse if we misunderstand it as a mere technology of automation or optimisation. An overemphasis on optimisation, paradoxically, can actively break down whatever coordination exists today, making the entire system less reliable.
Forget automation
Viewed as an automation tool, some parts of the organisation, with the necessary resources and urgency, improve their workflows and decision-making, while other parts, seeing less immediate returns on investment, continue to operate on outdated assumptions.
Exponential systems
Understanding exponential technologies but misunderstanding exponential systems. What matters isn’t the scale of processing power. What really matters are the cascading effects that emerge when coordination unlocks entirely new possibilities.
Framing the problem
We have a framing problem – a tendency to solve for a world whose underlying logic has already changed.
- AI is often seen as just another tool – something that enhances or replaces human effort without altering the fundamental structure of how knowledge, decision-making, or economic activity is organised.
- Executives view AI as a cost-cutting tool that can help them reduce headcount and improve profits.
- It assumes that jobs are stable units of work and that the primary effect of AI will be task-level substitution or enhancement.
AI doesn’t just augment or automate tasks; it changes how work is organised.
The things to consider
- AI treats jobs as stable bundles of tasks.
- Task-centric versus system-centric.
- It is not the skill, it is the coordination.
- Decoupling, reconfiguration, composability, or assembly will become increasingly precise.
- Automation versus augmentation is no longer relevant.
- Tacit versus explicit.
- The end of traditional pricing models.
- Representation determines reward. If the platform can’t measure it, it doesn’t get compensated.
- Economic value versus contextual value.
- Unbundling and rebundling.
- As AI learns and improves, it can change the location of power within an organisation.
- Is the customer interface the most valuable control point?
- Centralisation versus decentralisation.
- The value of nuance.
- Do you work above or below the algorithm?
- Where does value move when knowledge becomes abundant?
- The value of judgement. Judgment becomes important when AI-generated predictions and plausible answers are provided.
- Broad humans beat narrow AI.
- Curiosity, the ability to frame the right question before seeking answers, and curation, the discernment to elevate the most relevant answers for further action, become increasingly valuable as the ability to generate answers becomes commoditised by AI.
- Structure, when thoughtfully designed, doesn’t constrain freedom but dramatically expands it. Structure, paradoxically, unlocks team-level brilliance even in the absence of individual stars. Most organisations fail to design for it.
- The importance of algorithmic awareness.
- Capital vs labour.
- The importance of the management of constraints. Rebundling determines what is possible, but constraint management determines what is viable.
- The integration trap. The more you rely on someone else’s AI tool to power your business, the more control you hand over to the provider behind it.
- AI as a tool vs AI as an engine.
- Tool providers vs solution providers
- From outcomes-as-a-service to results as a service.
- Simplify decisions to build loyalty.
- Leverage the control point to rebundle the ecosystem.
- From customer journeys to integrated experiences
- Direct vs derived demand
Sommeliers
I love the bit in the book about sommeliers. A bottle of wine sells for $80 in stores. The restaurant charges $400. The diners think they’re paying for wine. What they’re actually buying is the sommelier; his vibe, his wit, his ability to make them feel like connoisseurs. The sommelier isn’t in the business of dishing information about wine. They’re in the curation business. Sommeliers rose to command greater value in a world where the product (wine) and the knowledge about it were getting commoditised. What the sommeliers figured out, before the rest of us eventually do, is that value is not in giving people more information. It’s in giving them confidence in a moment of uncertainty, making them feel like connoisseurs, even if this is the first sip they’ve ever had. The sommelier, it turns out, isn’t clinging to a dying profession. He’s living in the future. We just haven’t caught up yet.
Follow the constraint
The sommelier’s story is interesting, not because their original role remained untouched, but because it transformed to resolve the new constraint. Knowledge itself was restructured. The rise of elite credentialing bodies like the Court of Master Sommeliers turned the sommelier’s expertise into a visible, scarce signal. Don’t chase the skill, follow the constraint. What happened to the sommelier is now happening across the entire knowledge economy as AI transforms knowledge work.
Lego everywhere
AI enables agentic execution, making the execution of knowledge-intensive workflows also available as a building block. Leverage comes not just from combining available building blocks into new configurations, but also from creating your own building blocks by turning your unique skills into reusable components that others can leverage.
It is strategic
This is a fundamental integration question. Where are you in the value chain? And ultimately, who owns the data.
1. AI tools can absorb operational knowledge, particularly in manufacturing and process-intensive settings where machines equipped with sensors capture operational data.
2. AI tools may also absorb tacit knowledge and expertise that is held in the minds of skilled professionals or acquired through decades of practice.
3. As more companies adopt the same tool, any advantage the tool offers becomes a commodity available to the entire industry.
Be the tool
Are you the tool or the solution? There will be a growing power struggle between solution providers, who deliver value to end customers, and tool providers, who initially enable them but increasingly expand to dictate not just how solutions perform but also how entire industries operate. Tool providers secure power through unbundling and rebundling. They have the ongoing learning advantage. They have the clock speed.
Ownership
As AI tools rapidly improve, solution providers are redesigning their businesses to tap into the performance gains they offer. At first glance, this appears to be a win-win: the tool does its job, the solution becomes sharper, and customers are happier. However, the solution increasingly becomes dependent on an engine that the solution provider doesn’t control. When you’re building a solution where performance is fundamentally dictated by the tools you rely on, you need to think very seriously about taking ownership of those tools.
Ultimately, it will lead to results as a service
How far do you think tool providers will climb? Companies such as Formic, Winterhalter, Rolls-Royce, or Orica have shifted from selling tools to delivering solutions that guarantee performance, and this comes down to one thing: the ability to manage the risk associated with that guarantee.
Designing for indecision
I love this part of the book. Quantum leadership. Quantum is the new black. The law of unlimited possibilities. Where everything is full of potential until it is fixed. . Extreme unbundling until it is bundled (just-in-time bundling?).
The new map
It is pretty fundamental. Strategy is not about the road map. It is about creating a new map. Without a clear view of this shifting landscape, of what AI makes newly valuable or newly replaceable, it’s easy to misread the game that is being played. It is a whole different level of disruption, agility, and innovation capability. True strategic advantage in the age of AI doesn’t come from doing old things faster, but from understanding how the system itself is changing and deliberately reshaping it to your advantage.
Become a field reshaper
To play the winning game, you must first step outside the task, observe the new system forming, and then restructure the entire playing field. Become a field reshaper. Changing both the playing field and the rules of the game
