Can I AI?

Name a task. Answer five quick questions. Get a clear verdict (Automate, AI-Assist, or Keep Human) and watch it join a running map of every task you've sorted.

About the book behind this tool
Managing the Machine book cover

Managing the Machine is a practical, no-jargon guide written for the managers actually expected to make AI adoption work day to day. A recurring question in the book is a simple one: of everything on your team's plate, what actually belongs to a machine, what should a machine help with, and what has to stay with a person no matter how repetitive it looks?

This tool turns that question into a five-question scorecard. Score a task's Routineness (how repetitive it is, how clear the rules are) against its Human-Necessity (the stakes if it goes wrong, the context it depends on, how quietly a mistake could do damage), and you get a verdict grounded in the book's framework rather than a gut feeling.

Score one task, or score your whole task list. The workflow map below builds as you go.

View Managing the Machine on Amazon

Be specific: "screening resumes" scores differently than "final hiring decisions."

How this diagnostic works

Two scores, one verdict

Every task gets rated on five questions, 1–5 each. Questions 1–2 average into a Routineness Score (how repetitive it is, how clear and consistent the rules are). Questions 3–5 average into a Human-Necessity Score (the stakes if it's wrong, how much it depends on context an outsider wouldn't have, and how easily a mistake gets caught). Both scores are rescaled to 0–100.

The one veto

If you rate question 3 (the emotional, ethical, or reputational stakes of getting it wrong) at the top of the scale, the diagnostic stops and returns Keep Human immediately, skipping the remaining questions. No combination of routineness or convenience overrides a task where getting it wrong can do real, lasting harm.

Reading the grid

Outside the veto, a task lands in one of three zones: high Routineness paired with low Human-Necessity is Automate; low Routineness paired with high Human-Necessity is Keep Human; everything in between (which is most real tasks) is AI-Assist. There are no sub-tiers inside AI-Assist; it's one wide, deliberately unglamorous middle zone.

What this isn't

This is a starting judgment for one task at a time, not a compliance tool or an org-wide policy. Score enough tasks and the workflow map becomes a useful picture of where your team's automation effort is actually worth spending, but it's still a prompt for a conversation, not a verdict handed down from a machine.