Pyramid, flatter, loop.
What changed is not the org chart. It is how far a decision has to travel before it is made — and how much of that distance is covered by hand.
The way companies are drawn has changed twice in the last two centuries, and it is changing again. Read as a manifesto, this turns into a claim about the end of hierarchy. That is not what it is, and we are not making that claim.
Read as an operating question, it is narrow and practical: how many hands does a decision pass through before it is made, how many of those hand-offs are manual, and how long does it take for the result to come back to the place where the work happens?
Three shapes of the same question
The three shapes below are descriptions of how information moves, not stages a company is graded on. Most companies contain all three at once: one department runs as a loop while the one next to it still waits for a signature.
- 1840
Pyramid
Decisions are made at the top and travel down. Information travels back up at the speed of the people carrying it.
- 1990
Flatter
Fewer layers and a faster flow, but the hand-offs are still linear: each step waits for the one before it to finish and pass it on.
- 2030Outlook
Loop
Data, people and agents keep pace in one path: the signal, the decision and the record sit inside the flow instead of being passed along it.
The years mark the shape being described, not a measured transition. 2030 is a direction, not a deadline by which anything is due.
This is not the end of hierarchy. It is a shorter path from signal to decision.
This has been done once already, without software
There is one well-documented precedent for moving the decision closer to the work, and it did not come out of IT. From the 1950s onwards, Toyota Motor Corporation built what became known as the Toyota Production System, described later by its main architect, Taiichi Ohno. Two of its ideas are the relevant ones here.
The first is jidoka, and the andon cord that goes with it: the line is allowed to stop. A worker who sees a defect halts production instead of passing the problem to the next station. The signal is raised where the problem is, by the person who sees it, and it interrupts the flow immediately rather than surfacing in a report later.
The second is the pull signal, kanban. Instead of a central plan pushing parts down the line, each station signals upstream what it has just consumed. What gets produced is driven by what was actually taken, not by what somebody predicted would be needed.
Neither of these removed management from Toyota. The hierarchy stayed, and so did the plan. What changed was where information enters the system and how quickly it comes back — the loop was moved to the place where the work happens.
Source: the Toyota Production System as described by Taiichi Ohno in his own account of building it. Jidoka, andon, kanban and just-in-time are documented, publicly checkable concepts. We quote no figures here on purpose — the mechanism is the point, not a productivity number.
The same question, in an office
When an order arrives, how far does it travel before it is recorded? When a request needs an approval, how does the person who decides find out that it is waiting? When something goes wrong, does the signal reach the flow, or does it reach a report at the end of the month? These are questions about data paths, not about culture.
A loop, in this sense, is not a philosophy. It is a path: the signal is raised where the work is, the data that decision needs travels with it, the repeatable part is handled without a hand-off, and a person is brought in when it is an exception. That is what an operating layer is for. It does not flatten the company. It shortens the distance between something happening and somebody knowing.
Note on what is not in this text. It came out of a section that used to sit on the homepage. Two claims from that version are gone: that moving to the loop is not optional, and that whoever switches sooner gains a head start. Neither can be checked, so neither is published.