Most companies are asking the wrong AI question. They are asking: how do we add AI to the business? That sounds practical, but it's a trap.
The modern company was designed around human scarcity. Work was divided into functions because people had limited context. Layers of management existed because information moved slowly. Meetings existed because coordination was expensive. Software systems became systems of record because humans needed a shared place to store, retrieve, and approve work.
AI does not simply make that system faster. It changes the premise.
If one person can now prototype a product, generate customer research, write code, test workflows, produce creative variations, and coordinate agents across multiple tasks, then the basic unit of work is no longer the task. It is the workflow. And the basic unit of organization is no longer the department. It is the high-context pod — a small cross-functional team that owns a workflow end-to-end.
That is why so much "AI transformation" will disappoint. It assumes the company stays basically the same. The org chart remains. The process remains. The approval chain remains. The quarterly planning ritual remains. Then AI is inserted into each box as a productivity layer.
The big consulting firms have spent the last two years selling frameworks, maturity models, governance taxonomies, and centers of excellence — the entire apparatus of digital transformation, repackaged. Their own reports increasingly admit that the engagements aren't producing value. The proposed fix is always a larger engagement.
I'll be honest about my own position. Maginative builtAIMark, an eight-dimension maturity assessment model used by Fortune 500 leadership teams. So this critique cuts close. But any maturity framework built before the agentic shift — including ours — was implicitly measuring progress toward a more AI-enabled version of the company that already exists. That destination is no longer the right one. A useful framework today has to surface the structural question, not paper over it: are you trying to become a better version of yourself, or are you the wrong shape for what's coming?
Last year, Tobi Lütke, CEO of Shopify, shared amemothat he sent to the entire company:
Lütke was not announcing a tool rollout. He was changing the default operating assumption of the company: that the answer to a problem is more headcount. Once that default flips, almost everything downstream has to be rebuilt — performance reviews, hiring rubrics, planning cycles, team composition. Shopify made AI fluency a formal part of 360 reviews. Headcount has stayed roughly flat while revenue hascontinued to growat 20–40% a year. That is the redesign. Adding Copilot licenses is not.
Earlier this month, Brian Armstrong, CEO of Coinbase, went further. Hismemoannouncing the layoff of 14% of the company is the clearest statement yet of what an AI-native rebuild actually means in practice:
This is not an "AI strategy"; it's an organizational thesis. Armstrong is arguing that the unit of work has changed, so the unit of organization has to change with it. The layer cap, the manager redefinition, the one-person pods — those are second-order consequences of taking the first claim seriously.
Of course, you also don't need to look very hard to find examples of CEOs who failed. Klarna wentfurthest, fastest, on the rhetorical front — froze hiring, claimed its AI chatbot was doing the work of 700 humans, talked about cutting its workforce in half. Then the CEOwalked it back:"In a world of AI nothing will be as valuable as humans!"The lesson is not that AI-first is hype. It is that Klarna was solving the wrong problem. They treated AI as a way to simply replace people doing the existing work, rather than as a reason to redesign the work itself. The companies getting this right are asking what the work itself looks like when humans and agents are on the team together.
The important point is not layoffs. Some CEOs will absolutely use AI as a convenient story for cuts they wanted to make anyway. That will happen. But dismissing the whole pattern as cost-cutting misses the deeper shift.
The reality is that AI breaks the headcount logic of the firm.
For decades, growth meant adding people, then adding managers to coordinate those people, then adding systems to coordinate the managers. Complexity was the tax paid for scale. AI attacks that tax directly. It lets smaller groups absorb more context,