A Slogan Is Not a Diagnosis

Rejecting "AI Native" as the wrong word was easy. Anyone can do that in one blog post.

The harder question: what actually replaces it?

Not another word. Richard Rumelt already named this exact move in Good Strategy / Bad Strategy, and he named the fix too.

What "AI Native" actually is

Rumelt has a whole taxonomy of bad strategy. "AI Native" checks two boxes at once:

Fluff. Language dressed up enough to sound smart without saying anything that could turn out to be wrong. Try disagreeing with "AI Native" on the merits. You can't — there's nothing there to argue with.
A goal wearing a strategy's clothes. "Be the kind of company this platform was built for" is a destination, not a plan. It never names what's actually in the way.

Here's the part that matters more than the taxonomy: good strategy, Rumelt says, always starts with the kernel — three pieces, in order.

A diagnosis — the real nature of the challenge.
A guiding policy — the overall approach for dealing with what the diagnosis found.
Coherent actions — coordinated steps that carry the policy out.

Skip step one and steps two and three have nothing to be about. Which is exactly what "AI Native" does. It's step two, dressed up as the whole plan, with no diagnosis underneath it.

So what's the actual diagnosis?

Not "we're not native to this yet." That's not a diagnosis — it's the goal restated as if it explained itself.

Here's what I keep circling back to, for the shape of situation the last post described:

It's not a capability gap. Nobody's arguing the tools don't work.
Nobody's mapped the friction. Which decades-old, well-understood, high-volume workflows are genuinely good candidates? Which ones are judgment-heavy enough that automating them would be a mistake? A platform demo sells capability. It doesn't sell a map of where this specific company's capability gap actually is.
Review doesn't scale with output. Even where AI genuinely fits, someone still has to check the work, and checking is bounded by human attention in a way that adding more compute never touches. That constraint doesn't go away because the strategy has an exciting name.
The real asset gets treated like a liability. Decades of institutional knowledge, real client relationships, working systems — that's a pivot point a company born yesterday doesn't have and can't buy. "Native," as a word, is about what's new. It quietly asks you to look past the one thing that isn't.

That's the diagnosis. Not exciting. Actually true.

The guiding policy

A guiding policy isn't the plan. It's the direction that makes the diagnosis actionable.

If the obstacle is "we haven't mapped the friction and we haven't built the review capacity to check the work," the policy that follows is:

Find out what's already true before deciding what to buy.

Map the friction first. Size the review bottleneck before committing to the volume a platform promises to generate. Treat the institutional knowledge already in the building as the thing to point capability at — not the thing getting replaced by it.

Less exciting than "AI Native." Also the one you can actually act on.

What coherent action looks like

Not a platform rollout. Something narrower:

Pick one process that's genuinely well-understood and high-volume.
Prove it the way you'd prove any replacement — old and new running side by side until the diffs go quiet, not a vendor's whole platform on the strength of one demo.
Build the review step on purpose, sized to the volume this is actually going to produce, before that volume shows up.
Go slower than the six-month version everyone wants — because the six-month version was never a plan. It was a deadline with a goal taped to it. No diagnosis in between.

I still don't have this finished with the client.

What I have now is language for what's actually missing, instead of just a word I already knew was wrong.