The End of Delegation
For most of my career, one of the most normal management reflexes was forwarding.
A problem came in. You found the right person, the right function, the right layer. You added two lines of context and sent it across. Then you waited for the analysis, the note, the recommendation, the next meeting.
We called this delegation. Often, it was necessary. But over time, it also became a status marker. The more senior you became, the more you handed off. "I do not get into the details" was said with pride in many rooms, including rooms I was part of.
I think that is starting to invert.
Management was built for a world where context was scarce and coordination was expensive. One person could not easily hold the finance view, the customer view, the ops view, the product view and the legal view of a problem at the same time. So companies split the knowing across many people, then built structures to manage the split.
The first structure is visible. The ladder. CEO, CXO, VP, manager, executive. This is the hierarchy everyone talks about.
The second structure is less discussed, but just as powerful. The wall. Finance, HR, sales, operations, engineering. These started as skills. Then they became identities. People learned to say "I am a finance person" or "that is an ops problem" or "HR will handle this."
So the old company had two hierarchies. The horizontal hierarchy of designation. And the vertical hierarchy of function.
Flatland is our attempt to remove both. It is the operating system we are building at Cars24: one role, Builder; fewer titles and functional walls; open context; and ownership that stays with the person closest to the problem.
Yes, at Cars24 we removed titles, grades and bands. Everyone is a Builder now, including me. But if Flatland only removes designations, it is incomplete. The harder work is removing the walls.
A Builder may be working on a finance problem today. That does not mean their identity is finance. Tomorrow the same problem may need marketing, data, operations and product judgment together. In an AI-native company, the question cannot be: which function owns this? The question has to be: what does this problem need?
Function becomes a tool. It stops being an identity.
AI is what makes this possible. AI changes the cost of context. A Builder can now pull data, understand policy, model trade-offs, draft a PRD, inspect code, test flows and pressure-check assumptions much faster than before. Not perfectly. Not without judgment. Not as a replacement for deep expertise. But fast enough that the old reflex of throwing every problem over a wall starts to look weak.
This is the real inversion.
For years, delegation was proof that you had grown. In an AI-native company, "why are you delegating this?" becomes a serious question. Not because collaboration is bad. Not because people should work alone. But because handing off the understanding of a problem is very different from asking someone to help you solve it.
We are already seeing this inside Cars24. What used to feel like a top-100 operating layer is compressing toward roughly 50 real owners. I do not see that as a simple people statistic. I see it as the system telling us something important.
It does not need as many relay points anymore.
Work that earlier moved through senior coordination layers is now moving through fewer Builders who stay closer to the problem and use agents for leverage. The lesson is uncomfortable, but clear: what is losing value is not senior talent. What is losing value is seniority whose main job is delegation.
Football is the model I keep coming back to. Players pass the ball constantly. They coordinate, cover, create space and use each other's strengths. But nobody delegates the game. Nobody stands still and says goals belong to the striker's department. A pass is not an exit from ownership. It is part of the play.
That is how work should feel.
This does not mean expertise stops mattering. It means the expertise that compounds changes. Deep functional knowledge is still valuable. But the durable skill now is becoming AI-native: knowing how to gather context, direct agents, check their output, catch errors, move across domains and still own the outcome.
The old career advice was: pick a wall and climb the ladder inside it.
The new one is: go deep on AI-native work, then switch functions as the problem demands.
This matters more at Cars24 because we are not a pure software company. We work with physical cars, inspections, hubs, dealers, loans, RC transfers, service issues and customers whose problems are not theoretical. An AI-native operating model has to survive contact with real operations. It has to work where software meets the road.
That is harder. It is also why it is worth doing.
We are trying to build one of the first AI-native operating models for a real-world consumer business, from India. Not by adding AI as a tool on top of the old company, but by asking what parts of the company existed only because AI did not.
Inside Cars24, the direction is simple.
Delegate more tasks to agents.
Delegate fewer problems to each other.
When agents take away the routing, summarising, waiting and first-pass analysis, people do not become less important. The opposite happens. What remains is judgment, taste, relationships, standards and ownership. The human job gets bigger. But it gets bigger only if people stop acting like couriers of context.
I do not know exactly what the AI-native company will look like five years from now. Nobody does. But I am convinced the conditions that made delegation the default act of management have changed.
Management as delegation made sense when context was expensive.
We no longer live in that world.
And we should stop building companies as if we do.
Further reading: The complete Cars24 culture books on Flatland and Our Values.
Loved this article?
Hit the like button
Share this article
Spread the knowledge
More from the world of Cars24
Infosys’ Jayesh Sanghrajka is joining our Board
Bringing decades of experience in governance, public markets and scaling global businesses
Introducing Cars24 Labs: Our $20 million bet on AI
why we believe the next decade will belong to AI-native companies