Some of the least changed companies I know now call themselves AI-first. Often, they mean they have added AI to support, sales, coding, or search. The workflow may be better. The company itself may be much the same.

Cars24 began as an engineering company. As transactions grew, our attention moved toward operations: inspections, logistics, refurbishment, and financing. I let that happen. The company drifted from the reason it had started. Around 2023, it became clear that the old way would not take us forward. We had to change how the company worked while the live business kept running.

A founder I respect told me recently that his company was AI-first. I asked what had changed. He said they had integrated GPT into support. I asked what else. He paused, then said they were still figuring that out. I do not say this to mock him. Most of us built companies for a world where context was expensive to gather, share, and act on. That is the distinction I care about. An AI-first company uses AI. An AI-native company changes how it works because AI changes what is possible.

A company carries context

That can sound like a small difference until you look at what a company does. In 1937, economist Ronald Coase asked why firms exist instead of buying every task on the open market. He argued that using the market has costs. A firm can reduce them by coordinating work through managers instead. Hierarchy was not simply ego or bureaucracy. It was one way to gather information and coordinate action. Jay Galbraith later described organization design as a way to process information under uncertainty. Mel Conway showed how an organization's communication structure can shape the systems it builds.

A growing company gathers signals from the edges, makes sense of them, and turns them into decisions. Its structure affects who sees those signals, when they see them, and what gets lost along the way. As a company grows, no one person can see the whole system. Layers help summarize what is happening, pass important information upward, and carry decisions back down. That approach helped large organizations stay coordinated. It also added distance and delay.

A layer is a design choice

Every layer can add delay and reduce detail. Facts get polished as they move up. Decisions get simplified as they move down. By the time someone who can act sees the issue, the account may be cleaner but less useful than the reality that produced it.

AI is changing the cost of carrying context. It can summarize, route, sort, and draft an initial analysis. That does not replace judgment. It does mean some work once carried through a chain of people can happen faster and more continuously. The old structure deserves another look.

The question is not whether to use AI. Most companies will. The harder question is what kind of company makes sense when carrying context becomes cheaper. How many layers do we need? Which meetings exist mainly to move information? Which decisions wait because only a few people can see enough to make them? These are questions about how the company works, not which tools it buys. An AI-first company adds software to an old shape. An AI-native company asks whether that shape still serves the work.

In many companies, context travels in batches through meetings, decks, reviews, and escalation chains. People wait for the round trip before they act. Some recurring rituals are there because people do not share enough of the same picture. Status meetings, chase emails, escalation decks, and permission loops often try to solve that same problem. When people can see the situation together, meetings can focus on decisions instead of retelling what happened.

Our monthly P&L review is one example. It used to run for three hours. We began by waiting for someone from business finance to explain how revenue had moved and why the margin had shifted. Now people come in having already asked an agent to read the P&L and answer those questions. The diagnosis happens before the meeting. The meeting lasts half an hour, and most of it goes to what we do next and who owns it. The time in the room has shifted from establishing what happened to deciding what to do.

That is a useful change, but faster meetings alone do not make a company AI-native. If a year of work happens in a month, the business should also learn and improve faster. Otherwise, it may have produced more activity without more progress. I will not call Cars24 AI-native because of the tools we adopted. I will believe the change when we are visibly faster on what matters: customer experience, unit economics, growth, and how quickly we can do new things.

What changes, and what remains human

The lazy version of this argument says AI replaces managers. I do not think that is the right way to describe it. AI can reduce the coordination work around management: collecting status, reformatting updates for another layer, or meeting because no one can otherwise see the same picture. Judgment, hiring, standards, coaching, conflict, and hard calls under uncertainty still need people.

This does not mean every layer or form of authority should disappear. Structure can support accountability, incentives, quality, safety, and coherence. The narrower question is whether a layer still helps the work, or mainly relays information that people could share directly.

The change also depends on how people treat the truth. If people hoard context, soften bad news, or wait for permission because mistakes are punished more than delay, AI will not make the company wiser. It may help produce a polished account of the wrong story more quickly. Cheaper ways to carry context matter only if people can share what is really happening.

Shared context can give more people agency

Traditional organizations often move decisions upward because that is where information collects. An issue from a store goes to headquarters because headquarters has the wider picture. That can be a context problem rather than a character problem. AI may help by making more of the relevant picture available to the people closest to the issue. The goal is not to have more agents. It is to help more people see enough to act.

This is the change I have felt most directly at Cars24. What mattered was not simply adding tools. It was asking which coordination work still needed people and which could now happen directly, with shared context. Many companies will buy new tools and leave their old ways of working in place. They will summarize meetings that could disappear, make reports clearer while keeping decisions far from the work, and call that transformation. It may be useful modernization. It is not the same as changing the company.

Start with the shape

The shape of the company is part of its AI strategy. Start there and the questions change. Which meetings should go? Why does an update pass through three people before anyone acts? If you were building the company today, with AI available from the start, would you choose the same layers, rituals, and distance between signal and action?

If the answer is no, the gap between the company you have and the one you would build is the work of transformation. It reaches into authority, incentives, reporting rhythms, hiring, and status. That work can take years. Anyone who says a company became AI-native in six months should be able to show what changed in how it operates.

A McKinsey survey from 2025 tested 25 factors and reported that workflow redesign had the largest effect on whether companies saw a bottom-line impact from generative AI. That finding is one data point, not proof that every company should redesign in the same way. It does point to a useful question: what changed in the work after AI arrived?

The next decade will not split companies into those that use AI and those that do not. Most will use it. The difference will be between companies that keep their old shape and those that change how context moves, who can act, and how quickly they learn. That is the difference between AI-first and AI-native. One adopts the technology. The other redesigns the company around what the technology makes possible.

Notes and Sources