The AI-Native Shift · Book Resources
The One Shifts — all 17
One plain action per chapter, imperative and doable this week — each held to the stand-alone standard: you should be able to take the action and explain it to others without the book in hand, with no book-internal metaphor standing in for the instruction. Read as a set, they are the book’s action curriculum.
The One Everyone’s Afraid to Lose
Sit down this week with the one person your operation can’t run without. Have them correct one piece of AI output out loud, and write down the reason for every correction they make.
The Org Chart Is a Confession
Pull up your org chart and circle every place where work moves only because one specific person carries it by hand. Those names are your real structure — and your biggest risk.
The Complexity You’ve Normalized
Pick one workaround your team treats as normal — the double entry, the export step, the meeting that exists to fix a handoff. Document what it’s really costing you: the errors it causes, the customers it slows, the frustration it breeds — and what your people could be doing instead. Then decide, on purpose, whether to keep it.
Appetite Is Not Readiness
Find one process where AI output ships without a human judgment call. Rework that one this week: the machine produces the draft; a named person makes the call before it goes out.
What Done Looks Like
Pick one part of your operation and write three sentences: how it actually runs today, how it should run instead, and the name of the person whose knowledge gets you from one to the other.
The Meaning Layer
In your next customer meeting, count how many times the room talks about scores, stages, and pipelines — versus what the customer is trying to accomplish. Don’t fix anything yet. Just count.
Structure Follows Value
Follow one real customer relationship through your company and count every time it changes hands. For each handoff, ask: does this serve the customer, or just our org chart?
The Platform Question
Give your architecture document to someone two levels away from the design work and have them read it cold. If they can’t tell you what the company is building toward, the document isn’t done.
People Activate
Retire the word build from your next AI conversation. Instead, ask each person: what problem do you own that you’re sick of? Start there.
The Junction
The next time you correct an AI output, write the correction down as a rule with its reason — and store it where the system will apply it automatically, so that correction never has to be made twice.
The Small Death
When your best expert pushes back on AI output quality, don’t argue the quality. Have a private conversation about what’s actually at stake for them — and let them set their own pace.
Watch It Spread
Pick your most AI-capable person and ask: if they left tomorrow, would their replacement start where they finished — or from zero?
The Second Person
Find one correction your best person gave an AI this week, and ask where it went: into a shared system everyone benefits from, or into a private chat history that helps no one else.
Governance Worth Having
Find one place where routine, easily reversible work waits on someone’s signature. Remove that approval — and move the scrutiny to the actions that can’t be undone.
Counting What Counts
Take hours saved off next quarter’s board slide. Report two things instead: how much value each person created, and what the company can do now that it couldn’t do last quarter.
Handing Over the Keys
Ask the person who runs your most critical system: who would take this over if you left? If there’s no name, you’ve found the next thing to document.
Closing: Three Doorways, One Room
Don’t file this under someday. Pick your way in this week — even one conversation counts — and start.
The Closing’s shift is folded, unlabeled, into the coda by ruling. All seventeen were rewritten on 2026-07-04 to the stand-alone standard.