For two hundred years, strategy was bottlenecked by the cost of knowing things. Research was slow. Analysis was expensive. The firm that could gather more, model more, and read more had an edge — because most couldn't afford to.
That bottleneck is gone. Analysis now costs roughly nothing. Anyone with a prompt can produce a competitor teardown, a market scan, and five scenarios before lunch — work that used to take a team a quarter.
Most people are responding by getting faster at the thing that no longer matters. They're optimizing the part the machine just commoditized. The advantage didn't disappear — it moved. It moved to the one place AI can't follow.
Let me show you where, because the line is sharper than the hype admits.
Strategy is four moves.
Strip strategy down and it's the same loop every time, whether you call it scenario planning or just running a business. I think of it as four moves — Scan, Story, Stake, Steer. It's the working shape of the Wack–Chermack scenario tradition I came up in, and it's also just what you actually do when the plan meets the world.
- Scan — read the signals. What's moving, what's noise.
- Story — turn the signals into a few coherent futures you could plausibly be living in.
- Stake — commit. Place the bet, with your name on it.
- Steer — adjust as the real world votes.
AI changed two of these completely and didn't touch the other two. That gap is the whole story.
Scan: this is what AI is for.
The Scan is pattern-finding at scale, and it's exactly what these models are best at. Feed it the landscape and it will surface what you'd have missed, faster than any analyst. This is not a threat. This is the best research assistant ever built, and you should use it without guilt or limit.
If your edge was gathering — being the most informed person in the room — that edge is now available to everyone for the price of a subscription. Being informed is table stakes. It is no longer a position.
Story: AI drafts, you commit.
The Story move is where it gets interesting. AI will happily generate scenarios — ten of them, articulate and plausible. That's useful. But notice what it's doing: it's producing options, not conviction. It will give you every future with equal confidence and no preference, because it has no preference. It has nothing riding on which one is true.
Choosing which story you're going to bet the company on — that's not analysis. That's taste. The model can hand you the menu. It cannot tell you what you want for dinner, because it isn't the one eating.
Stake: the part that can't be outsourced.
Here's the line. The Stake is the moment of accountability — the irreducibly human one.
To stake something is to commit resources to a future that hasn't happened yet, knowing you'll be wrong some of the time, and to own the consequences when you are. It requires skin in the game. The model has none. It will never be fired, never lose the capital, never have to look the team in the eye after the bet goes sideways. It is, structurally, incapable of having a stake — and a decision without a stake isn't a decision. It's a suggestion.
"The AI told us to" is not a strategy. Strategy is the willingness to be held responsible for a choice under uncertainty.
You can automate the reasoning that leads up to the choice. You cannot automate the standing behind it. The accountability is the job.
Steer: judgment in motion.
And once you've staked, the world starts voting. Steer is reading the live feedback and knowing — not calculating, knowing — when a wobble is noise to ride out and when it's the signal that your story was wrong and the bet needs to change. AI can monitor the dashboard. It can't feel the difference between a dip and a turn, because feeling the difference is built from having been burned before.
The real danger isn't the robots.
The failure mode I'd watch for isn't that AI replaces strategists. It's subtler, and already happening: teams outsource the analysis to the machine, and then quietly outsource the judgment along with it. They let the confident, fluent output stand in for a decision nobody actually made.
The result is strategy that is articulate, well-researched, internally consistent — and ownerless. Nobody's name is on it. Nobody will defend it in eighteen months. It reads like a plan and behaves like a hedge. When the market tests it, there's no one home.
That's the thing to protect against. Not the tool — the abdication.
The job moved up the stack.
So no, the strategist's job didn't shrink in the age of AI. It moved. It moved off the Scan, which is now free, and up onto the Stake, which is now the only scarce thing left. Your value is no longer how much you can analyze. It's the quality of the bets you're willing to put your name on, and your nerve to steer them in public.
I write this from inside the work, not above it. I run six operating companies. Every framework I publish gets staked on my own balance sheet before it reaches anyone else's — and AI runs my scan every single day. It's extraordinary at it. But it has never once made a decision for me, because it has nothing to lose if it's wrong, and I do.
That's not a limitation of the technology. That's the definition of the job.
— Tim