You're looking at AI from the wrong angle

Over the past few months, I’ve consumed an absurd amount of content about AI. Articles, newsletters, podcasts, reports. Every day something new appears. A broken benchmark, a more powerful model, a company that laid off half its staff, another that tripled its revenue. The volume of information is so high that you start thinking you’ve already understood the landscape.

I thought I had. I’ve written here before about productivity, about the risk of outsourcing your thinking, about how most companies are reacting to hype instead of looking at data. I thought my perspective was reasonably well-informed. Then I started reading Reshuffle by Sangeet Paul Choudary.

Right in the first chapter, something made me stop. It wasn’t exactly new information. It was a reframing. A different way of looking at something I thought I already understood. And when that happens, everything that follows changes color.

The wrong conversation

Most discussions about AI revolve around one word: automation. “AI will replace lawyers.” “AI will eliminate office work.” “AI will do what 10 people used to do.” Substitution. Speed. Efficiency. It’s always about doing what already exists, just faster and cheaper.

Choudary proposes something different. His central thesis is that AI’s real economic impact doesn’t come from automating tasks, but from coordinating fragmented systems. AI is not, in its most transformative sense, a tool that does things for you. It’s an infrastructure that allows dispersed actors to work together in ways that were previously unviable.

It sounds subtle. It’s not.

Coordination as competitive advantage

Think about the most valuable companies in the world. Uber doesn’t own cars. Airbnb doesn’t own properties. Amazon started without inventory. What these companies do is coordinate. Connect those who have something with those who need something, at scale, with reliability.

Choudary calls this “the means of coordination.” He argues that, historically, the biggest economic shifts didn’t come from new production tools. They came from new forms of coordination. The modern corporation, for example, didn’t emerge because someone invented better machines. It emerged because someone solved the problem of coordinating specialized work under one roof.

AI, in his view, is the next great coordination revolution. Not because it makes individuals more productive (it does that too), but because it drastically reduces the cost of aligning incentives, integrating information, and making different systems talk to each other. Things that previously required hierarchies, contracts, entire bureaucratic processes.

The argument goes further: AI doesn’t just reduce existing coordination costs. It unlocks coordination that simply didn’t exist before. Problems nobody tried to solve because the cost of aligning all parties was too high.

The analogy that convinced me

Of all the points in the first chapter, the one that struck me most was the shipping container analogy.

Before the standardized container, maritime trade existed. Ships carried cargo. Ports operated. But the process was slow, expensive, and full of friction. Each port had its own loading and unloading system. Each ship was different. The logistics of moving goods between countries was a nightmare of incompatibilities.

The container didn’t make ships faster. It didn’t invent maritime transport. What it did was standardize the interface. Suddenly, any product could be loaded onto any ship, unloaded at any port, placed on any truck. The container made global trade reliable.

Choudary argues that AI does something analogous for knowledge work. It doesn’t make people smarter. It makes collaboration between people, teams, and systems scalable and reliable. AI is the shipping container of intellectual work.

This analogy changed how I think about the subject. Because if AI is about coordination, not automation, the questions that matter are completely different.

If it’s not automation, what changes?

When the frame is automation, the questions are predictable. “Who loses their job?” “What task does AI do better?” “How do I become irreplaceable?” These are defensive questions. Focused on individual protection.

When the frame is coordination, the questions shift. “What kind of collaboration was impossible before and is now viable?” “What problems went unsolved because they required aligning too many parties?” “What markets didn’t exist because the cost of connecting supply and demand was prohibitive?”

It’s not “who loses their job.” It’s “who connects better.”

Choudary points out that the greatest value gains in the economy don’t come from making existing processes more efficient. They come from unlocking new value flows that weren’t previously possible. AI as coordinator opens exactly this type of opportunity: it doesn’t optimize the current game, it creates new ones.

And this has a profound implication for anyone working in technology, strategy, or any kind of knowledge work. The competitive advantage isn’t in who uses AI to do things faster. It’s in who uses AI to connect what was disconnected.

Rethinking my own experience

I use AI every day. I’ve talked about it more than once. Claude for writing code, reviewing architecture, structuring arguments. The individual productivity gain is real and significant.

But after reading the first chapter of Reshuffle, I started seeing my own usage differently.

Most of what I do with AI is, honestly, sophisticated automation. I take a task I’d do alone, use the tool, and do it faster. It’s valuable. But it’s the most obvious version of what the technology can do.

The moments when AI truly transformed my work were different. They were when it allowed me to coordinate things I wouldn’t have coordinated on my own. Integrating technical contexts with business contexts at a speed that made cross-functional conversations viable. Exploring five architecture alternatives simultaneously that, without the tool, I would have done sequentially (and probably stopped at the second). Connecting information from different sources I would never have placed side by side manually.

These aren’t speed gains. They’re coordination gains. And the difference, now, feels enormous.

The operating system of collective action

Choudary uses an expression that stuck with me: AI as the “operating system of collective action.” The idea is that, just as an operating system coordinates hardware, memory, and applications so the user doesn’t need to think about each component separately, AI can coordinate people, data, and processes so that organizations can function in ways that were previously too complex.

This goes far beyond “me + tool = more productivity.” It’s about what AI enables groups of people to do together. It’s about reducing friction between parties that need to collaborate but historically ran into coordination costs, misaligned incentives, or simply a lack of visibility into what the other side needed.

If this thesis is correct (and the arguments are compelling), the question we should ask ourselves isn’t “what does AI do for me.” It’s “what does AI allow us to do together.”

What I’m taking from this

I’m still early in the book. But the first chapter already changed the lens through which I evaluate AI’s impact.

Not that automation doesn’t matter. It does, a lot. Individual productivity gains are real, measurable, and will keep growing. But if we stay stuck in that frame, we miss what might be the bigger transformation: the ability to coordinate work, knowledge, and decisions at a scale that was never possible before.

AI isn’t just a more powerful calculator. It’s a new way of organizing collective action. And whoever understands this first will have an advantage that can’t be measured in tasks per hour.

---

If this topic interests you, I’d love to exchange ideas. Find me on LinkedIn.