Doing the Same Thing with AI Isn't Transformation

A few weeks ago, I wrote about how Reshuffle by Sangeet Paul Choudary changed my perspective on AI. The central thesis of the early chapters was that AI’s transformative impact comes not from automating tasks, but from coordinating fragmented systems. It’s not about doing things faster. It’s about connecting what was disconnected.

I kept reading. And chapter 3 deepened something I see happening in the day-to-day of almost every company: the temptation to fit AI into existing activities and call it innovation.

The optimization trap

Most companies I know are using AI the same way: taking existing processes and speeding them up. The report that took three hours now takes twenty minutes. Email triage got faster. Customer service got a chatbot.

These are real gains. But Choudary makes an important provocation: if you adopt AI without changing how coordination works, you’re building leverage for the tool provider, not for yourself.

Read that again. If the only thing that changes is speed, the advantage isn’t yours. It belongs to whoever sold you the tool. Because any competitor can buy the same tool and reach the same speed. The difference lies in who uses the technology to rethink how the pieces connect.

Task efficiency vs. system reconfiguration

Choudary separates two types of AI gains. The first is task efficiency: doing what you already did, just faster. The second is system reconfiguration: changing how decisions flow, how people coordinate, how value is created.

The most significant gains don’t come from faster execution. They come from changing how decisions flow through the system and how new decision points create new centers of relevance.

For leaders, this completely changes the question. It’s not “where can I deploy AI?” It’s “do the activities we do still make sense the way they’re designed?”

The example that made me think about education

Here’s an example that came to mind reading chapter 3, thinking about education (a sector I know closely).

A school can use AI to grade exams faster. That’s task efficiency. The teacher saves time. Grades come out sooner. Everyone’s happy.

But the reconfiguration question is different: if the student has access to the same AI as the teacher, what does “assessment” even mean? What exactly does a traditional exam test when the student can ask AI to solve it? The entire format of evaluation needs to be rethought, not just accelerated.

The school that only uses AI to optimize the existing process is doing the same thing faster. The school that rethinks what “assessment” means in a world with AI is reconfiguring the system. The second creates an advantage the first will never reach, no matter how fast it grades exams.

Coordination isn’t just for giants

An important point from chapter 3: AI allows smaller companies to achieve levels of coordination that previously required enormous scale. You no longer need traditional market power to orchestrate an ecosystem.

This is relevant for anyone leading operations of any size. A small business that uses AI to coordinate suppliers, partners, and customers in a new way can create a competitive position that doesn’t depend on size. It depends on how the parts connect.

Choudary puts it this way: the more essential your system becomes to how others coordinate, the less substitutable you are. It’s not about being the biggest. It’s about being the most necessary within a network of relationships.

The question that stays

After reading this chapter, I started looking at my own work differently. How many of the things I do with AI are truly reconfiguration? And how many are just the accelerated version of what I was already doing?

Being honest: most of it is still optimization. And there’s nothing wrong with that, the productivity gains are real. But if I stop there, I’m only capturing the most obvious part of what the technology offers.

The question for leaders isn’t “does the company already use AI?” It’s “are the activities the company does still the right activities?”

Because if the structures have changed and we keep doing the same things (just faster), it’s not transformation. It’s acceleration of what already existed. And in a world where the entire system is reconfiguring, optimizing the old game might be the most elegant way to fall behind.

---

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