It’s not every day that a business strategy essay reshapes an industry, but Jack Dorsey and Roelof Botha’s From Hierarchy to Intelligence is having an unusual run. Published in March, it laid out a vision of AI-era organizations — companies built as intelligences rather than hierarchies, with thin layers of edge workers replacing traditional management. Two months later, Coinbase CEO Brian Armstrong announced layoffs and explicitly adopted the vision, down to its vocabulary.
Block already laid off 4,000 people — 40% of its workforce. The essay is Dorsey’s justification. But does it stand up to scrutiny?
The Vision
The vision rests on two models and a layer that combines them.
The first is a company world model — a continuously updated representation of operations built from the artifacts of remote-first work (Slack messages, design docs, decisions). The model shows what’s being built, what’s blocked, where resources are allocated, what’s working and what isn’t.
The second is a customer world model — a per-customer, per-merchant representation built from transaction data. Block sees both sides of millions of transactions daily through Square (the merchant) and Cash App (the buyer). Dorsey and Botha argue this is an unusually clean signal: People lie on surveys. But when they spend, that’s the truth.
Between these models sits the intelligence layer, which composes financial primitives — payments, lending, card issuance, banking — into customer solutions on-demand. The example: a restaurant whose cash flow is tightening before a seasonal dip. The intelligence layer composes a short-term loan, adjusts the repayment schedule, and surfaces it to the merchant before they even think to look for financing. No product manager decided to build the offer. The capabilities existed; the intelligence layer recognized the moment and composed them.
What disappears in this picture is the traditional product roadmap and the management layer that maintained it. The org normalizes to three roles: Individual Contributors who build capabilities and the model itself, Directly Responsible Individuals who own cross-cutting customer outcomes, and player-coaches who combine building with developing people. There is no need for a permanent middle management layer.
What’s Actually New
When the intelligence layer tries to compose a solution and can’t because the capability doesn’t exist, that failure signal is the future roadmap.— Dorsey & Botha, From Hierarchy to Intelligence
Dynamic, data-driven financial product offers aren’t new. The infrastructure to evaluate creditworthiness in real time, compose a specific offer, and surface it to a customer is well-established. Block already does versions of this through Square Capital. So what’s the vision actually claiming?
The vision’s claim is that the organization that builds these offers is itself unnecessary. Existing fintech composes dynamic offers through product teams: managers, roadmaps, quarterly planning, specific features designed and tested deliberately. The claim is that this organizational infrastructure can be replaced. The intelligence layer composes offers from primitives based on the customer model. No product manager decided to build a solution. The capabilities existed. The intelligence layer recognized the moment and composed them.
Block has an advantage here: fintech has a relatively narrow product space. Whether through regulation, financial modeling, or the underlying mathematics of money movement, the degrees of freedom are constrained. There are only so many ways to structure a loan, a payment, or a card. A consumer social product has nothing like this constraint — the space of what could be built is open-ended, and figuring out what’s worth building is most of the work. For Block, the composition problem is genuinely well-suited to the vision.
When DRIs Collide
A DRI might own the problem of merchant churn in a specific segment for 90 days, with full authority to pull resources from the world model team, the lending capability team, and the interface team as needed.— Dorsey & Botha, From Hierarchy to Intelligence
If composition is the new product work, the DRI is the new product manager. The vision gives them broad authority and tight time horizons. “Full authority to pull resources” sounds clear until you ask what happens when two DRIs need the same engineer. In an org that’s deliberately flat, there’s no shared manager to escalate to. The vision is smart not to burden AI with decision-making, because after all, AIs are not wise. Yet somebody needs to resolve conflicts like this.
Then there’s the experience of being a DRI. Ninety days on a problem, then move to the next one. No permanent team. No colleagues you’ve worked with for years who’ll fight for you when resources get tight. You’re constantly negotiating across organizational boundaries you have no enduring relationship with.
Some people thrive in this kind of work. They have high political skill, comfort with conflict, and don’t need long-term stakes in specific outcomes to feel grounded. But these aren’t most people. And they’re not obviously the same people the vision describes as needed at the edge — those with “intuition, opinionated direction, cultural context, trust dynamics, the feeling in a room.” It seems like those traits would develop better through a sustained relationship with work and people. Steve Jobs’ thoughts on consulting come to mind.
The vision’s answer is that the world model holds enough institutional knowledge to substitute for what permanent teams used to provide. But there’s a difference between information being stored in a model and someone’s lived experience.
Pour one out for the middle manager
There is no need for a permanent middle management layer. Everything else the old hierarchy did, the system coordinates, and everyone is empowered, with a role that’s much closer to the work and the customer.— Dorsey & Botha, From Hierarchy to Intelligence
The vision treats middle management as information routing — the function that AI now handles. Block goes from 10,000 employees to 6,000 partly by removing middle layers the World Model is supposed to replace.
Player coaches will be responsible for developing people, but middle managers also resolved conflicts, absorbed risk for their teams, and made local decisions. The middle manager who fights for their team, who earns their trust, and who, in return, commands a loyal squad ready to tackle any given challenge is a huge asset to any company.
Authority is part of what makes these relationships work. When a manager has the power to advocate for you — to push back on a bad deadline, to fight for a promotion, to take a risk on your behalf — that’s when trust deepens into loyalty. Strip out the authority and the manager becomes another peer with opinions. The relationship loses what made it valuable. If we’re at our best when we’re working for each other, an Intelligence that flattens those bonds into coordination protocols is missing something.
AI isn’t the enemy
What does your company understand that is genuinely hard to understand, and is that understanding getting deeper every day? If the answer is nothing, AI is just a cost optimization story. You cut headcount, improve margins for a few quarters, and eventually get absorbed by something smarter.— Dorsey & Botha, From Hierarchy to Intelligence
Dorsey and Botha definitely have an enemy in mind. The essay is pitched as a once-in-a-lifetime AI opportunity, but it’s also a response to the threat of AI labs like Anthropic. Elsewhere Dorsey has spoken about the “existential dread” he felt when considering the potential of AI.
The financial picture supports this reading. Block’s $1.3B in net income on $24B revenue is a modest 5.4% margin — normal for fintech, but not the position of a company that can afford complacency. Removing roughly $1B in annual costs through the cuts could nearly double net income while signaling to competitors that Block can operate lean.
Read this way, the vision is partly a defensive posture. Radical cost reduction, removal of organizational layers, redirection of effort toward building AI infrastructure before the AI labs do — this is a company that thinks it might not have a future if it doesn’t act fast.
Conclusion
As organizations grow into the thousands, they revert to hierarchical coordination because no alternative information routing mechanism has been powerful enough to replace it.— Dorsey & Botha, From Hierarchy to Intelligence
The essay opens with the Roman Army and traces hierarchy through Prussian staff officers, American railroads, and Frederick Taylor’s scientific management. The framing is rhetorically powerful: hierarchy is information routing, and AI is better information routing. If you accept the analogy, the conclusion mostly follows.
But isn’t hierarchy more than just a framework for message passing? It grants authority to resolve conflicts, it maps a path for ambition, and provides a stable environment for collaboration and learning. The essay asks us to believe these are either bugs to be removed or features the world model can replicate.
Hierarchy is reduced to information routing. Management is reduced to coordination. Product development is reduced to composition. Business itself is reduced to “understanding.” The essay names Zappos, Spotify, and Valve as previous attempts to escape hierarchy — each of which reverted as it scaled. Block’s story has a new technology and a new framework, but the underlying confidence is familiar. It lets executives make hard decisions while believing they’re not making trade-offs. The middle managers being let go aren’t a loss; they’re overhead. The product teams being dissolved weren’t doing real work; they were coordination tax.
But not all valuable knowledge fits on a graph. Some of it lives in relationships that took years to build, in judgment developed through experience. The vision treats this knowledge as either capturable by the world model or as residual edge work the system can’t reach yet. This underestimates what’s actually being lost. It’s a different kind of company entirely, and we don’t yet know if it’s one that can sustain itself.