The org chart is the bottleneck

When electricity replaced steam, manufacturers pulled out the steam engine and bolted an electric motor into the same spot.

They kept the shafts and the belts and the whole layout, because the layout was designed around one big source of power in the middle of the building and nobody had yet asked why that still had to be true. It took about thirty years before someone put a small motor on each machine, tore out the belts, and redesigned the floor around the flow of work rather than the flow of power.

We're in that thirty year gap.

The claim

Reorganising the org chart around AI-era workflows matters more right now than picking which AI tools or models a team adopts.

The technology is ready. The gains from bolting AI onto an existing workflow are real, and they are capped by the structure they're plugged into, since a faster motor on the same belt still turns the belt at the speed the belt allows. Most of the energy in the industry is going into the motor anyway: which model, which assistant, which vendor, how many seats. Those are the easy arguments to have, because they're arguments about procurement, and procurement doesn't threaten anybody.

Why it's political, not technical

Processes exist because someone designed them, someone owns them, and someone's authority depends on them.

That is the whole reason the org chart is the bottleneck and the tooling isn't. Swapping one assistant for another costs a budget line. Removing a handoff costs somebody a role, or a meeting they chair, or a sign-off that made them relevant, and every step in the chain has a person attached who would (quite reasonably) prefer it stayed. Conway's law says the system you build mirrors the communication structure of the organisation that builds it. Point AI at an org built for human handoffs and you get human handoffs, faster.

The belts are still there.

I don't take reorgs lightly. The bar I've written about before is one sentence you have to be able to say without flinching: the current structure is the thing preventing us from doing the work. For most orgs trying to go AI native, that sentence is now true, and most of them haven't said it yet.

Reliance is the wrong question

What usually stands in for a redesign is a strategy meeting about how much to rely on AI.

Good question. Wrong one.

Your nervous system handles about eleven million bits a second and your conscious mind about fifty. The body never holds a strategy meeting about appropriate reliance levels. It routes. Reflexes at the spinal cord, pattern recognition in the basal ganglia, novel decisions in the prefrontal cortex. So the thing worth asking is what architecture lets each kind of intelligence do what it is actually suited for. Reflexes for the predictable, attention for the novel.

Most software orgs are still stuck in the strategy meeting, treating AI as a vendor they procure and allow in through a gate. It keeps producing the same result: a carefully governed layer of AI-assisted work sitting on top of an organisation that hasn't otherwise changed. That's the motor in the old engine's spot, with a committee deciding how fast it's allowed to spin.

The organism includes its symbionts

About forty trillion microorganisms live in a human body, roughly equal to human cells by count. We don't say "the human plus its bacteria." We say the human.

The organism includes its symbionts. The ones that work with it are what make it viable, and the ones that try to sterilise the environment tend not to survive.

Organisations still drawing a hard line between "our team" and "the AI" are fighting their own biology, and you can see the line on the org chart if you look for it: the AI initiative as a separate programme, with its own owner and its own governance, bolted to the side of the org rather than routed through it. Where the analogy breaks, and it's worth naming, is that a nervous system doesn't have to persuade cells to participate.

Organisations do. No tooling solves that, which is one more reason the hard part lives in the org chart and not the procurement list.

What leaders have to give up

Redesigning the floor means letting go of the things that made the old floor legible. Three, in particular.

The first is the task as the unit of accountability. Tickets closed, pull requests merged and story points burned all rise while value delivered stays flat, because they track activity inside a process rather than what the process produces. Adoption numbers are the newest member of that family.

Second, the interface as proof of work. The standup, the status update and the weekly report exist partly for coordination and partly because they let managers observe effort, and when the system routes work and surfaces only exceptions, that visible surface shrinks. Leaders who can't read outcomes without process theatre will find this deeply uncomfortable. The discomfort is diagnostic rather than a reason to slow down.

And the AI initiative as a project with an end date. Substitution projects close, while ambient ones compound as the system accumulates context, so funding it like a deployment means cutting resources exactly as the returns start to accelerate. The substitution trap is partly a funding model.

Each of the three is a way of seeing the work that depends on the work being done in a particular structure. Change the structure and you lose the view, and it feels like losing control even when nothing has gone wrong.

Where to start

Not with a new org chart drawn on a whiteboard. That's the other failure, and just as expensive. Start by finding the belts. Trace one feature from idea to production and count every human handoff: PM to design, design to engineering, engineering to QA, QA back to engineering, engineering to staging, staging to production. Each one adds latency, miscommunication and context loss. Then give people both the tools and the explicit permission to cross role boundaries, because the permission is usually the missing half, and then ask what each step is for if the outcome could be reached without it.

The answers will name people. That's the political part, and it's the part no vendor can sell you. It's also why the tooling conversation is so comfortable. Delivery is the loop everyone can see, so it gets the new motor, while the structure around it keeps turning at the old speed.

The technology is ready. The floor isn't.