Earning Speed
… or, why mise en place matters even more with agentic AI.
When my son Anthony came home from his culinary arts program he turned into “Kitchen Cop”. I was not thrilled. He had professional kitchen experience, a ServSafe certificate, and very little patience for the way we had always done things. He started correcting where things went, how they got cleaned, when surfaces got wiped down, how raw food got handled, and what had to be ready before we even turned on the stove. At first, it felt like a hassle. Then something changed. As I got used to it, the whole kitchen got faster. Dishes stopped piling up. Prep stopped crashing into cleanup. Cooking got easier and fun again. Nice little protocols like announcing when you’re holding something hot or walking behind someone made us move like we were choreographed.
That’s mise en place. In cooking, it means “putting in place.” In plain English, it means getting things ready before the real work starts and executing with a plan. Why? Because speed isn’t something you demand in the middle of action; speed is something you earn before action starts. You earn it by getting ready, narrowing the work, cutting extra decisions, and refusing the kinds of mess that slow you down later.
You don’t move fast because you’re in a hurry, you move fast because you already did the work that makes hurry possible.
Most people think speed comes from urgency. They think fast teams are fast because they push harder, decide faster, and put up with more chaos. That can work for a short time, but then the bill shows up. Real speed comes from earlier discipline: clean tools, clear scope, good boundaries, fewer surprises, and fewer hidden problems waiting to blow up. By the time the “fast” moment arrives, the team isn’t making it up as it goes. It’s cashing in work it did earlier.
The Trap
This is also true in software. A team that says yes to every feature request, every exception, every edge case, every new tool, and every little convenience may look flexible in the moment, But it’s often building a slower future one missed opportunity at a time. David Heinemeier Hansson says this again and again when he says things like constraints are liberating, saying no is productive, and low-cost clean systems preserve optionality. This is execution design. You move fast later because you refused the wrong things earlier.
Get David Bock’s stories in your inbox
Join Medium for free to get updates from this writer.
Nate B Jones says the same thing from the AI side when he argues that agents fail not because the model is weak, but because the team handed it work that was never prepared well. Missing context, weak boundaries, bad decomposition, and vague success checks can look like model problems from a distance. Up close, they’re often mise en place problems. The work wasn’t shaped into a form that a person or a machine could carry safely. So the system improvises, and then leaders blame the tool.
Earned Speed
That’s why the phrase earned speed matters. It puts the focus back where it belongs. Not on fake urgency. Not on velocity theater. Not on “move fast” as a slogan. And not on tools that are optimized for time-to-first-demo. Earned speed asks a better question: what happened upstream? What complexity got refused? What context got prepared? What assumptions got made clear? What constraints got written down? What did the team leave out on purpose?
Bryan Liles says something close to this when he argues that we’re losing the conditions where judgment forms. AI compresses time and compresses human count. Work that once took several people over several days can now get pushed through by one person in an afternoon. That’s real acceleration. But it also strips away some of the time and social contact where understanding used to grow. That means the need for earned speed is rising at the exact moment the conditions for earning it are getting thinner. If the old world used to hide discipline, reflection, and correction inside ordinary work, the new world doesn’t. You have to design for them on purpose.
Charity Majors says the same thing when she argues that if AI makes it easier to generate change, then rigor has to move closer to consequences. That’s another way of saying speed has to be earned. You don’t get to generate recklessly and then hope reality is kind. You need tighter feedback loops, better observability, closer contact with the real system, and stronger human understanding. Otherwise you aren’t moving fast. You’re just hitting failure sooner.
Mise en place belongs at the center of this conversation. It gives a name to something many teams feel but don’t describe well. Preparation isn’t work in addition to the work, Preparation is part of the work. Clean context is part of the work. Negotiation is part of the work. Documentation is part of the work. Good teams negotiate scope and tradeoffs early enough that refusal stays clean and execution stays fast. The person who says no to the wrong complexity helps create speed. The person who lays out the context packet, the test surface, the eval boundary, the tools, the constraints, and the cleanup path also helps create speed. The person who keeps the sink empty enough for the next move to be easy is helping create speed.
What Teams Get Wrong
Many organizations don’t want to hear this. They want speed without the disciplines that create it. They want output without preparation. They want acceleration without refusal. They want the AI dividend without the judgment infrastructure. But that’s not how it works. You don’t move fast because you’re in a hurry. You move fast because you already did the work that makes hurry possible.
If real speed has to be earned through preparation, refusal, and contact with reality, then teams that fall in love with things that only look fast are not building real leverage. They’re getting better at making unfinished work look complete.
