Ed SchaeferEd Schaefer

AI-Native Ways of Working: A Leader's Self-Assessment

Sixteen quick questions. About five minutes. No login, no email required — just an honest read of where things actually stand today.

Most leaders have a gut sense of how AI-assisted work is really going on their team. This turns that gut sense into a number you can act on. Answer based on what's actually happening right now, not what you're planning or hoping for. There are no wrong answers, and no one sees your results but you.

How this works

Sixteen statements, grouped into four short sections. For each one, rate how often it's true for your team right now, using the same scale for every statement:

ScoreHow often
0Rarely or never
1Sometimes
2Usually
3Basically always

By "AI tool" we mean whichever AI your team uses to do or check real work — pick the one that matters most if you use several.

Add up your score at the end of each section. Add the four section totals for your score out of 48. Find your band at the end.

Your answers stay in this browser tab. Nothing is sent anywhere, and nothing is saved once you leave the page.

Section 1: Autonomy & Oversight

1.When an AI tool is about to do something outside its usual lane, someone actually stops and checks first — instead of just letting it go ahead.
2.People on your team feel safe admitting they used AI on something, and just as safe admitting they didn't understand what it gave them.
3.When your team and an AI tool might both start the same piece of work, someone follows a clear rule for who takes it — not just whoever notices first.
4.When AI-assisted work starts piling up, someone actually enforces a real limit, so your best people still get time to think, not just review and unblock things.

Section 2: Checking the Work

5.When AI-built work gets handed off, someone actually walks through what it does, out loud — not just points at the fact that it "looks done."
6.Before AI-assisted work counts as finished, a person or a separate process actually checks it.
7.When someone asks whether AI is actually helping, you point to a real outcome number — decisions here don't get made on vibes. (A count of AI-generated PRs doesn't count on its own; score this on whether the number tracks an outcome, not activity.)
8.When an AI tool does something wrong, people know what to do next without scrambling.

Section 3: Learning & Growth

9.When an AI tool makes the same mistake twice, someone fixes it for good, with a written rule or check, instead of explaining it again next time.
10.A person defines the problem and what "done" means before an AI tool starts building it.
11.When working with AI tools gets confusing or frustrating, someone specific actually steps in and helps your team adjust.
12.When AI could easily do work that would otherwise teach a newer or junior person something, someone still makes sure they get real practice building that judgment.

Section 4: Foundation & Leadership

13.When someone asks what your AI tools have done recently, you actually pull up a record — not just rely on whoever happened to be watching.
14.Someone actually keeps what your AI tools need to know up to date (the docs, wikis, or context files your AI tools are pointed at — sometimes called context engineering), so nobody's constantly re-explaining the same background from scratch.
15.Your leadership team actually reviews how the whole AI-and-people setup is working, not just what got shipped.
16.You personally use the AI tools your team relies on often enough — at least once in the last month — that your calls about them come from firsthand experience, not just what you've been told.

0 of 16 rated. Finish rating every statement to see your score and band.

Updates

New essays, by email, as they publish.

A guided session with someone outside your team usually moves this faster than working it out alone.

Talk about the Diagnostic