AI Will Break Command and Control

The CEO becomes the bottleneck

Imagine a company where the CEO reviews requests and features, keeps track of what the product and engineering teams are doing and gets involved in a lot of the detail.

As the speed of development gets faster, the number of reviews and decisions the CEO has to make goes up as well. That will slow them down. The CEO becomes the bottleneck and essentially, you get fucked.

The 10x test

A good test today: imagine if your engineering team could do 10 times more than they currently do. Well, they probably can. What would you do?

I think it's always been a good test for scaling in the companies that I joined. Imagine you had 10x the people or 10x the output. It's a good test of where the bottlenecks are and where things break.

Let's say you have one product team. What changes if it's 10 product teams?

Same with capacity. You have to adjust your communication and leadership style. Instead of relying on verbal communication, more of it has to be written because that's the only way to scale. Not cancelling verbal. Verbal still makes sense where applicable, but the main workflow goes into the written.

The founder trap

It's also about founders, not only CEOs. A founder can be the CEO, but founders have a lot of context because they've been involved from the beginning of the business.

They might feel like, "Oh, it's my baby. I need to be involved in this."

That's when the founder really needs to realise it and break this pattern, because it's actually dangerous.

Even if you like being in the room and being the smartest guy or smartest person there, your role has changed. Your initial job might have been to go and sell and hustle the business, but now you need to invest in people and spend more time with them.

Pass that knowledge. Ask questions. Be more like a coach, even if it feels uncomfortable.

If you want to improve throughput and get faster, that's what you need to do.

What should leadership still be involved in?

If your engineering team has that capacity, what would they do? Do you have to be involved in all 10 features?

The best way to probably structure it is to ask: what are the bigger strategic items that you really care about, potentially associated with additional spend, risk or exposure? You should be involved in those.

It's also about whether it's a one-way door, where it's hard to reverse the decision.

You can use the rocks, pebbles and sand concept. The big rocks are the things you probably want to be involved in. But otherwise, use the strategic lens and let the team know what's important.

Typically through OKRsObjectives and Key ResultsA goal-setting framework that pairs a clear objective with measurable results showing whether progress was made.Learn more or whatever framework you use.

But let's say it's optimising a particular metric and the team knows quite a bit about the customer, technology and data. Let them fucking make decisions and experiment.

Why this gets dangerous

Typically, this sort of leadership style is hard to change and the CEO will struggle.

Companies that keep operating like this will be fucked or will have to change their leadership, because other companies that give teams context and intent, enable them to do their best job and set clear goals and strategy will win.

They'll have more autonomy and make decisions much faster without looking back at the leadership team and asking for permission.

We just need to remember that even with AI and all the tools, from coding to consolidating customer feedback to highlighting what's important and writing OKRs for the teams, it doesn't automatically make companies faster.

Yes, it will increase the amount of work they can do, but it will also obviously increase the cognitive load.

Whether it helps you move faster in the right direction and build the right things depends on how quickly you can make these decisions.

Command and control worked better when execution was expensive and slow, just because the stakes were high.

But now the game has changed. You've got to move fast and get to production. You can test your ideas pretty quickly.