
For you
Role Checks
Module · What goes to AI, what stays human
The Manager's AI Check
A manager's core act is delegation, and AI just added a second place to delegate to. Get the split right and you free your people for work that grows them; get it wrong and you either waste them on grind or quietly hollow out the judgment they need to develop. This module checks how you decide what goes to AI versus a person, how hard you review what comes back, and whether you are protecting your team's growth while you automate the busywork.
What the five levels look like
Every dimension in this assessment is scored 1 to 5. This is what the levels mean, dimension by dimension. The graded report diagnoses where your own answers land and what to do about it.
You know what goes where
- 1No logic
- 2Whatever is faster
- 3Rough rules
- 4Deliberate criteria
- 5Criteria your team knows
At the low end: Routing work with no logic means you are optimising nothing and can defend nothing. Write down what makes a task right for AI and what makes it right for a person, then use it. What good looks like: Deliberate, shared criteria are what make your delegation fair and legible. Keep them visible to the team so they understand why work lands where it does, and can push back.
You review AI-assisted work
- 1Wave it through
- 2Trust the tool
- 3Spot checks
- 4Review the substance
- 5Review and coach
At the low end: Waving AI-assisted work through because it looks clean is how confident errors reach customers. Review the substance of what ships, not the polish, exactly as you would human work. What good looks like: Reviewing substance and coaching from it turns each review into a lesson your team keeps. Keep it up; the review is where your people learn to catch what the tool gets wrong.
You protect their growth
- 1Never considered
- 2Let it slide
- 3Occasional stretch work
- 4Protect hard problems
- 5Grow them deliberately
At the low end: If you have never considered this, your team may be getting faster and shallower at the same time. Identify the tasks where their judgment actually develops and make sure AI does not quietly take them all. What good looks like: Deliberately growing your people while using AI is the hardest balance in the job, and you are holding it. Keep watching which tasks build judgment; the answer shifts as the tools improve.
You automate the busywork
- 1Everything manual
- 2Talk about it
- 3One or two tries
- 4Several automated
- 5Ruthlessly trimmed
At the low end: Keeping every report and meeting manual taxes your whole team by default. Pick one recurring status or summary this week and let AI draft it, then keep only the human parts that add judgment. What good looks like: Ruthlessly trimming the busywork is how you protect your team's attention for real work. Keep pruning; new routine reporting accumulates as fast as you remove the old.
Your team has norms
- 1No norms
- 2Everyone improvises
- 3Loose guidance
- 4Agreed norms
- 5Norms lived daily
At the low end: With no norms, your team's AI use is a patchwork of private guesses. Agree a short set of expectations, what to use it for, what to disclose, what to check, so nobody has to invent the rules alone. What good looks like: Norms your team actually lives by are what make AI use consistent and honest. Revisit them as the tools and the work change; norms set once and forgotten drift out of date fast.