What About Teams?

The recent From Hierarchy to Intelligence piece by Jack Dorsey and Roelof Botha landed on a nerve for a reason.

It is not just another “AI will make companies faster” argument. It is more structural than that. It asks whether hierarchy itself was mostly a workaround for one enduring constraint: humans do not coordinate well at scale.

That framing is powerful. It is also incomplete.

The obvious reaction is to focus on managers, org charts, and information flow. But the more interesting question sits one layer lower:

If hierarchy weakens, what happens to teams?

Teams were never the point

We often talk about teams as if they exist because collaboration is inherently good. As if someone, at some point, simply decided that people should work together and everyone nodded.

Historically, teams existed because coordination was limited.

A manager could only handle so many people. A group could only carry so much shared context. Communication had to be bounded. So we created teams — not because they were ideal, but because brains do not scale horizontally.

A team was a unit of coordination under human cognitive limits, a constraint I explored earlier in The Cognitive Ceiling: Am I Hitting Mine?, where cognitive load, not hierarchy, quietly defines how much work can actually be coordinated.

That definition shaped everything: team size, ownership boundaries, communication patterns, and even what we call alignment.

AI breaks the original reason teams exist

The core claim in Sequoia’s piece is that coordination is no longer human-bottlenecked.

A system can maintain a real-time model of the company, connect people to problems directly, and route information without relying on layers of managers to move context up and down the chain.

If that holds, then a lot of traditional assumptions begin to wobble.

• Teams are no longer required for information routing.
• Managers are no longer required for context aggregation.
• Alignment no longer has to depend on recurring meetings. (Someone tell my calendar.)

That leads to an uncomfortable implication:

The classic team is, at least partially, an artifact of a fading constraint.

So do teams disappear?

Not quite.

This is where the clean model starts to crack. And clean models always crack — usually around the part where actual humans show up.

Teams were never just coordination units. They were also trust networks, context containers, identity anchors, and learning environments.

More importantly, they are often where the whole becomes greater than the parts.

The missing variable: emergence

There is something important that does not show up well in system diagrams.

Call it emergence.

In a strong team, shared context compounds. Trust removes friction. Ideas build instead of restarting. Energy becomes contagious.

I explored a version of this in The Fantastic 4: when a very small group really clicks, it behaves less like four individuals and more like a single thinking system.

A system can coordinate.

It does not automatically create that. No algorithm yet produces the feeling of finishing each other’s architectural sentences.

Why individuals plus system is not the same thing as a team

A fully intelligent organization sounds compelling on paper.

Work finds the best individual. Context is globally accessible. Coordination is automated.

But several deeply human things remain stubborn.

Trust does not scale cleanly

Trust takes time. It requires repeated interaction and shared history.

Even with perfect information, people still ask a very analog question:

Do I trust how this person thinks?

That question is not solved by access to more artifacts. You cannot Ctrl+F your way to trusting someone’s judgment.

Engagement is local

People rarely feel motivated by a company graph or a globally optimized routing system.

They feel motivated by their people, shared momentum, and a sense that they are building something with others they know well.

A system can assign work.

It does not easily create belonging. Nobody has ever said “I stayed at that company because the task router was incredible.”

Creativity often prefers familiarity

Small, stable groups develop shorthand, shared taste, and faster iteration loops.

That is hard to reproduce in constantly reshuffled environments, even if the reshuffling is algorithmically efficient.

The constraint did not disappear

Even if coordination scales better, relationships still do not scale infinitely.

This is part of why the Sequoia argument is so provocative: it attacks a real bottleneck, but not the whole human reality around it.

Hierarchy was one workaround for human limits. Removing hierarchy does not remove those limits themselves.

And then reality shows up

This is the part most teamless visions tend to underplay.

Humans are not always available.

People get sick. They go on vacation. They burn out. They lose context. They switch more slowly than systems assume. Sometimes they just have a really bad Wednesday.

In a purely DRI-driven, system-routed model, an obvious question appears immediately:

What happens when the best person for the problem is unavailable?

You get fragility.

Redundancy is not optional

High-performing teams often create shared ownership, implicit backups, and overlapping context.

Not because that is maximally efficient on paper, but because it is resilient.

A system can route tasks very well. It does not automatically guarantee continuity, coverage, or graceful degradation.

Bus factor is still a human problem

We have always had some version of the same concern: what happens if one key person disappears?

Teams mitigate that through knowledge sharing, pairing, and shared accountability.

A highly individualized model pulls in the opposite direction. It tends toward hyper-specialization and single-threaded ownership.

Efficient, until it is not.

A more realistic model: three layers

Instead of framing the future as teams versus no teams, a more grounded model starts to emerge.

1. Stable trust layer

Small, consistent groups that build trust, share context, provide belonging, and create redundancy.

This is where engagement lives. It is where emergence happens. It is where resilience starts.

2. Dynamic execution layer

Work flows across boundaries. Problems assemble people. Coordination becomes more system-driven.

This is where efficiency improves and silos weaken.

3. Explicit resilience layer

Backup ownership, shared knowledge, and overlapping capability.

Historically, teams handled much of this implicitly. In a more system-first organization, it has to be designed on purpose.

Teams do not disappear, they evolve

Traditional teams bundled everything together: coordination, execution, trust, resilience, and identity.

That bundle now starts to split.

Function	Old world	Emerging model
Coordination	Team + manager	System
Execution	Team	Dynamic groups
Trust	Team	Team
Identity	Team	Team
Resilience	Implicit	Explicit design

Teams lose their monopoly on execution.

They remain critical for almost everything else humans care about.

A better question

The future question is probably not:

How should we structure teams?

It is closer to this:

How do we combine system efficiency with human reality?

That is a harder and more interesting design problem. Also, arguably, a more honest one.

Toward an adjusted model

The Dorsey vision is useful because it forces a reset. It takes seriously the possibility that systems can handle much of what hierarchy historically did.

That part matters.

But the stronger model is not hierarchy or teamlessness. It is a merge between system-coordinated work and human-shaped teams.

1. Empowered small teams as a home base

Stable, trusted, context-rich, resilient groups.

Not necessarily the only unit of execution, but the place where belonging, taste, and mutual confidence compound.

2. Dynamic work across teams

Fluid execution, temporary problem-based groupings, and system-routed collaboration that cuts across formal boundaries.

3. Built-in redundancy

No single-threaded ownership for critical areas. Intentional overlap. Shared understanding. Backup paths that survive real life.

The synthesis

Not hierarchy.

Not teamless.

Something more human.

Systems optimize how work flows.
Teams optimize how humans operate.
Redundancy helps the system survive reality.

Closing thought

The shift is often framed as hierarchy to intelligence.

For teams, it may look more like this instead:

team -> trust network + system + resilience

Because even in a perfectly coordinated system, people still get sick, lose context, need trust, and look for belonging.

The organizations that win will probably not be the ones that remove teams entirely.

They will be the ones that preserve what makes teams powerful while removing what made them necessary in the first place.