Somewhere in the next eighteen months, somebody is going to walk into your conference room with a technology nobody present understands and a strong opinion that you need it immediately. Maybe it runs a process end to end. Maybe it is something that does not have a name yet. It will arrive the way the last one did, with a vendor deck, a competitor who supposedly already has it, and a board member who read an article on an airplane.

The question is not whether you will handle that one well. The question is whether you have anything in place to handle it at all, or whether you will do what most organizations do, which is improvise under pressure and call it strategy.

We watch this from close range. An organization decides in a single meeting to restructure around a tool nobody has evaluated against anything. It is the same instinct as telling your contractor to get the roofer out here Thursday because he has some genuinely impressive roofing equipment. You do not have a foundation yet. It does not matter how good the equipment is.

* * *

Here is the part worth sitting with. The organizations that handled AI badly are not bad at AI. Their problem is more general than that.

They have no standing process for absorbing a technology. So every one arrives as an emergency, gets handled by whoever is loudest or most frightened, and produces a decision that has to be walked back a year later at considerable expense. Then the next technology arrives and the whole thing runs again from the beginning, because nothing was built that would still be there when it did.

There is a reason aviation does not work that way. You don’t invent the response to an engine failure during the engine failure. Every emergency procedure is written, trained, and drilled long before the day it is needed. Not because the procedures are clever. Because they exist in advance of the moment when your judgment is at its worst. The entire value of an emergency procedure is that it was built when nobody was scared.

Most organizations have no emergency procedures for technology. They have enthusiasm, a budget, and a competitor doing something.

* * *

Ed and I used to teach our young copilots to wind the clock before responding to something bad that happened in flight. Older aircraft had eight-day mechanical clocks on the panel that had to be spring-wound by hand. It took a few seconds, and it had nothing to do with flying, but it slowed you down and let you catch your breath.

That was the point. The few seconds it took to wind the clock put your head back in the airplane. Stop being excited about everything happening around you. Pause. Look around. Decide what’s really happening. Then fly the plane and deal with the emergency.

Leaders facing a new technology need an equivalent, and most of them do not have one. There is no institutional moment where somebody is required to stop and ask what problem this is supposed to solve before the money moves.

* * *

What the discipline actually looks like

None of this is complicated. It is a cadence, and the reason it works is that it exists before you need it.

Refresh the strategy on a schedule. Not when something frightening shows up in the market. Annually, at minimum. Review it, revise what needs revising, and be honest about what has gone stale, because it goes stale faster than it used to. Come out of that review with objectives and, underneath them, the actual tasks that have to be accomplished to reach them. Most strategy documents stop before the tasks, which is why they cannot be used for anything.

Build the structure that executes it. Oriented around the products and services you actually deliver, with support functions supporting rather than obstructing. Roles, responsibilities, and the relationships between those roles, which is the half almost everybody skips and the half that prevents silos. And explicit decision authority. Who decides what, at what level, against what criteria.

Determine capability and capacity against the tasks. What capabilities does the work require, and how much of each. Drive that off the tasks, not off what you happen to have on the payroll today. Then compare it to what you have. The difference is your gap, and it is the only honest input to any technology conversation.

Now ask the technology question. Does this close a gap, widen one, or neither? That question is unanswerable without the first three steps, which is exactly why so many organizations skip straight to it and end up buying something impressive that solves a problem they do not have.

Then build a transition plan. Where the organization is today. Where it needs to be in five years. The roadmap between the two, with the sequence written down and the decision points named. Almost nobody does this. They make the purchase, announce the change, and leave the middle to sort itself out.

* * *

There is a failure mode worth naming here, because it is seductive and it looks like progress.

Watch the humanoid robots run a race. If the actual objective were to get something around a track quickly, the solution would not look like a person. It would have wheels. The humanoid shape is not there because it is the best solution to the problem. It is there because it is impressive, and because it resembles the thing we already recognize. Then it runs straight into the wall.

Organizations do this constantly. They adopt the shape of a solution instead of solving the problem. They buy the tool the competitor announced, or the one that demonstrates well, or the one that most resembles what a modern company is supposed to have. The gap analysis is what prevents it. If you know which capability you are short, you can tell the difference between a tool that closes it and a tool that photographs well.

One more thing belongs in that gap analysis, and it is the one nobody counts. Some of the capabilities you will need in five years are the ones you are currently deciding not to grow. If the work that used to develop your next generation of experts gets handed to a tool, the tool will be fine and the pipeline will not be. Count that as a gap, because it is one, and it is the slowest kind to close.

* * *

None of this is an argument against the technology. We use AI in our own work every day. A diagram that used to take one of us fifteen hours takes about ninety seconds now, and then an hour to fix what it got wrong, which is still a trade worth making every time.

So the advice we give leaders is not to be careful. It is to be disciplined, which is a different thing and considerably more useful. Do not be frightened of what is coming. It is a tool, like every tool before it, and it will be followed by another one soon enough. But figure out how to use it before you bring it in, and build the process that makes that figuring routine instead of heroic.

Wind the clock. Get your head back in the airplane. Then decide.

Mike Westman and Ed Gillcrist are co-owners and co-CEOs of Shackleton Group, Inc., and the authors of the forthcoming Suits, Guitars & 3D Glasses: A Practical Guide to Corporate Maneuver Warfare®.