What Did Block Just Publish?
On March 31, Jack Dorsey and Sequoia’s Roelof Botha published an essay called “From Hierarchy to Intelligence.” The thesis is simple and radical: the corporate hierarchy that has organized human work for 2,000 years exists because humans were the only available coordination mechanism. AI changes that. Permanently.
Block isn’t bolting AI onto their org chart. They’re replacing the org chart. They cut roughly 4,000 positions, nearly 40% of their workforce, and restructured the entire company around AI systems that perform the coordination functions middle management used to handle.
This isn’t a cost-cutting exercise dressed up in theory. It’s a structural bet that the coordination layer of every company is about to be automated, and that the humans who remain become dramatically more valuable because of it.
We’ve been building exactly this pattern for small businesses since day one. Block just gave it a name.
Why Has Hierarchy Lasted 2,000 Years?
The essay traces management structure back to the Roman legion. Eight soldiers report to one leader. Eighty to the next. Four hundred eighty to the next. Five thousand to a general. Span of control: three to eight people. That ratio has barely changed in two millennia.
The Prussian military invented middle management after getting destroyed by Napoleon. Staff officers whose entire job was routing information between the people doing the work and the people making the decisions. Railroads copied the model. Factories scaled it. Taylor measured it. McKinsey productized it. Every company you’ve ever worked for runs on some version of this 200-year-old idea.
And the fundamental tradeoff has never been solved: add layers to coordinate more people, and you slow down information flow. Remove layers to speed up, and you exceed human span of control. Every “flat org” experiment (Spotify’s squads, Zappos’ Holacracy, Valve’s open allocation) eventually reverted to hierarchy because nothing else could handle the coordination load.
Until now.
What Is Block Actually Replacing?
Dorsey and Botha make a distinction that most AI coverage misses. They’re not giving employees AI copilots. They’re not adding chatbots to Slack. They’re replacing what hierarchy does: routing information, maintaining alignment, precomputing decisions. AI systems that perform those functions continuously.
The architecture has four components:
Capabilities. Atomic building blocks (payments, lending, card issuance, payroll) that have no UI and no opinion about how they’re used. Just reliability and compliance.
World Models. Two of them. One tracks everything happening inside the company: what’s being built, what’s blocked, where resources are allocated. The other maps customer and merchant behavior from transaction data. Together, they replace the contextual awareness that managers used to carry in their heads.
Intelligence Layer. Composes capabilities into solutions based on what the world models see. A merchant’s cash flow is tightening ahead of a seasonal dip? The system composes a short-term loan with adjusted repayment before the merchant even asks. No product manager hypothesized that feature. The customer’s reality generated it.
Interfaces. Square, Cash App, Afterpay. Delivery surfaces. The value lives in the models and intelligence layer, not in the UI.
What remains are three human roles: individual contributors who build the system, directly responsible individuals who own specific outcomes for 90-day cycles, and player-coaches who write code while developing people. No permanent middle management layer. The system coordinates.
Why Does This Matter If You Have 10 Employees?
Because you already are the coordination layer.
In a 10-person company, the owner is the world model, the intelligence layer, and the middle manager all at once. You’re the one who knows that the new client’s project is behind because the developer is waiting on assets from the designer who’s out sick. You’re the one who remembers that last quarter’s revenue dipped in March and you need to push invoices early this year. You carry the entire context of the business in your head, and you spend half your day routing it to the people who need it.
Block formalized what happens when you replace that coordination burden with AI. Here’s what it looks like at your scale:
Your “world model” is an AI system that knows your clients, your projects, your cash flow, your team’s capacity, and your calendar. Not a dashboard you check. A system that acts on what it sees.
Your “intelligence layer” is the AI that composes actions from that context. Client hasn’t paid in 30 days? The system sends the follow-up, adjusts the project timeline, and flags the risk to you. New lead came in from the website? The system qualifies them, checks your capacity, and books the call. Or tells them you’re full until next month.
Your people move to the edge. They stop attending status meetings, stop updating spreadsheets, stop chasing approvals. They do the work that requires judgment, creativity, and trust. The stuff that makes them the bottleneck the whole system depends on.
The pattern is the same whether you’re running a $40 billion fintech company or a 10-person service business. Replace coordination with intelligence. Keep humans where they’re irreplaceable.
What Block Gets Right That Everyone Else Misses
Most companies are using AI as a productivity tool. Write emails faster, summarize documents, generate code. Dorsey calls this “giving everyone a copilot” and argues it misses the point entirely. Copilots don’t change the structure. They make the existing structure slightly faster while preserving every bottleneck, every approval chain, every information relay that slows the organization down.
The structural move is harder and more valuable: ask what the hierarchy actually does, then build systems that do it better.
Block asks a question that every business owner should steal:
What does your company understand that is genuinely hard to understand, and is that understanding getting deeper every day? “From Hierarchy to Intelligence,” Block & Sequoia Capital
If the answer is nothing, if your AI is just autocomplete on top of generic processes, then it’s a cost optimization tool. Useful, but temporary. Someone with deeper understanding will absorb you.
If the answer is something specific (you understand your customers’ seasonal patterns, your market’s pricing dynamics, your service area’s regulatory quirks) then AI becomes the thing that compounds that understanding into a permanent advantage.
For small businesses, this is the entire game. You don’t have Block’s transaction data or Square’s merchant network. But you have something they never will: deep, specific knowledge of your 200 clients, your local market, and the exact problems you solve. That knowledge, fed into AI systems that learn from every interaction, creates a compounding advantage that no competitor can replicate without living your business.
What Block Gets Wrong
The essay is compelling, but it sidesteps a few things worth naming.
The timing is suspicious. Publishing a manifesto about AI-native organizations one month after cutting 4,000 people invites cynicism. Current and former Block employees have pushed back, noting that AI-generated code still requires significant human review and that regulated domains like banking aren’t ready for autonomous AI coordination. The theory is sound. The execution is unproven at this scale.
World models are fragile. The essay assumes clean, machine-readable data flowing through a remote-first company where all work creates artifacts. Most businesses, especially small ones, don’t generate clean data. Conversations happen on phone calls. Decisions happen in parking lots. Knowledge lives in one person’s head. Building a world model for a 10-person plumbing company is a fundamentally different problem than building one for Block, and the essay doesn’t acknowledge that.
The human edge isn’t just “edge cases.” Block positions humans as sensors who “reach into places the model can’t go yet,” as if the goal is to eventually automate them too. We take a different position. Humans aren’t the temporary gap in the model. They’re the permanent bottleneck that captures all the value. The point isn’t to build AI that eventually replaces human judgment. The point is to build AI that makes human judgment worth more.
What This Means for Your Business
You don’t need to read a 5,000-word essay from a billionaire to apply this. The pattern is three moves:
1. Identify your coordination tax. How many hours per week do you spend routing information between people, following up on tasks, and maintaining context that could live in a system? For most small business owners, it’s 15–25 hours. That’s your coordination layer.
2. Build your world model. Not Block’s version. Yours. An AI system that knows your clients, your pipeline, your team’s workload, and your business rules. It doesn’t need to be perfect on day one. It needs to start learning.
3. Move yourself to the edge. Stop being the router. Start being the decision-maker, the relationship-builder, the person whose judgment the whole system depends on. That’s where the value compounds.
Block wrote the theory. We build the implementation. Same thesis, different scale. And for most businesses, the smaller version is the one that actually ships.
The Standard
For 2,000 years, hierarchy was the only technology that could coordinate human work at scale. That era is ending. Not because AI is smarter than humans, but because AI can do the coordination work that humans were never meant to do in the first place.
You are no longer the engine. You are the grid.
The question isn’t whether your business needs this transition. It’s whether you’ll design it yourself or have it designed for you. Take our AI Readiness Assessment to find out where your coordination layer is costing you the most.
