The standard line on AI and strategy is that it's an accelerator — it does the things you were already doing, faster. That's true, and it's also an incomplete way to think about what's actually changing.
The more significant shift isn't speed. It's that AI makes things viable that simply weren't viable before. A team that couldn't have justified building its own internal tool a year ago can plausibly build one now. A product team that iterated in quarters can plausibly iterate in weeks. An idea that would have died at "someone should really look into that" can now actually get looked into, built, and shipped, by one person, in an afternoon.
On its own, that's a genuine gain. The problem shows up one step later.
More options isn't the same as more progress
When a new option becomes viable, a team has to actively decide not to pursue it. That sounds trivial, but it's a different kind of decision than teams are used to making, and a harder one.
Previously, a lot of prioritisation happened by accident. Most ideas never became real candidates for a team's time, because the cost of acting on them was high enough to filter them out automatically. Nobody had to make a deliberate call that the internal tool wasn't worth building — it was never going to get built anyway, so the question never came up. Scarcity was doing the prioritisation work for you, quietly, in the background.
AI removes a lot of that scarcity. Which means it also removes the free, accidental filtering that scarcity provided. Every option that used to get silently screened out now has to be screened out on purpose, by someone, making an actual judgment call about relevance.
And that's the part that makes this harder than it sounds: everyone already knows how to spot a pointless meeting or a vanity project. The harder case is the one where a team builds something impressive, well-executed, and genuinely useful, and it's still the wrong thing to have spent the quarter on. A sharp new internal dashboard, a clever product experiment, a fast iteration on an interesting idea — none of that reads as waste in the moment. It reads as momentum. That's exactly what makes it dangerous to a strategy: it's stimulating enough, and good enough, that nobody feels the need to ask whether it was ever actually the priority.
This is a different problem than "AI can write your OKRs"
I've written before about why AI doesn't fix the real bottleneck in OKRs — the mechanics of writing and tracking objectives were never the hard part, so a tool that writes them faster doesn't touch the actual failure mode. This is close to the opposite kind of claim.
That earlier problem shows up after a team has already decided what to work on, when it's time to write it down honestly. This one shows up before that — at the point where a team is deciding what even counts as a candidate for its attention in the first place. AI doesn't make that earlier decision harder because it offers less. It makes it harder because it offers more, and more of it is legitimately good.
Why this makes the strategy map more valuable, not less
I saw this play out directly with a client — a family-owned grocery business whose leadership team had a real tendency to take on whatever new idea had the most energy behind it in the room. Once that team had an actual weighted strategy map and had gone through prioritisation scoring together, something specific changed: they started asking, before taking on a new initiative, whether it actually connected to one of their real strategic priorities — and if it did, what would have to stop to make room for it.
That second question is the one most teams skip, with or without AI involved. It's also the one that matters more as the number of plausible new things to do keeps climbing. A team with a real, weighted view of its own priorities can look at an exciting new option and ask a concrete question: does this serve what we already said mattered most, and what are we willing to give up to do it? A team without that view just accumulates initiatives, each one individually defensible, collectively pulling in no particular direction.
This is, I think, the actual relationship between AI and strategy work — not that AI makes strategy obsolete, and not that AI is some kind of distraction to be managed. It's that AI raises the price of not having done the strategy work properly in the first place. The teams who already have a clear, genuinely prioritised view of where their effort should go are positioned to treat AI exactly the way it's usually pitched: as acceleration. The teams who don't are going to experience the exact same capability as scattering — more things happening, more energy in the room, and no clearer a picture of whether any of it mattered a year from now.
The tools didn't create the discipline problem. They just made it a lot more expensive to not have the discipline.