Skip to “Try This” for the AI prompt.
You hand off a responsibility, explain the steps, and make yourself available for questions. A week later the work comes back technically complete, but something important is missing. You can see it almost immediately; explaining what you saw takes much longer. So you revise it yourself, and the responsibility drifts back to your desk.
What’s usually missing isn’t the process. It’s the judgment behind it: which details deserve attention, when an exception matters, what makes something ready to move forward. Experience makes those distinctions so familiar that you forget they need explaining.
The Part You Leave Out
Imagine a director handing proposal reviews to a capable project manager with clear instructions: check the scope, review the schedule, confirm the fee, flag concerns. The project manager follows them carefully, and the proposals still come back for revisions. What the director never said is that “review the schedule” means noticing whether a deadline depends on information the client hasn’t committed to providing, or that “check the scope” means catching language that leaves ownership of a deliverable unclear. The checklist names the category while the reasoning stays in the director’s head.
Let AI Interview You
Putting that reasoning into words from a blank page is hard. One useful approach is to have AI interview you about a responsibility you know well, anchored in a real example. The example is what makes it work. Ask AI how an experienced leader reviews a proposal and you’ll get generic advice; describe a specific review, what you noticed and why it mattered, and the conversation has something concrete to work with. (Use an approved tool and remove anything confidential first.)
The AI supplies questions and structure. You still have to recall the situation, explain your reasoning, and decide whether the result reflects how you actually think.
Try This (Prompt Included)
Paste this into ChatGPT, Claude, Gemini, or similar, and fill in the brackets.
Help me explain the judgment behind a responsibility I’m handing to a colleague.
Responsibility: [describe]
Colleague’s relevant experience: [describe briefly]
Anonymized recent example: [the work, what I noticed, what I decided]
Interview me before writing anything. Ask one question at a time, up to six, and use each answer to choose the next. Explore what I check first and why, what makes me pause, how I tell acceptable work from work needing revision, exceptions that change my approach, and which decisions the colleague can make alone versus bring to me. Ask for a concrete example when I’m vague.
Then draft a concise handover guide with: (1) the intended outcome; (2) what to notice, why it matters, and what action to consider; (3) one worked example drawn only from what I shared; (4) decision boundaries and open questions.
Separate my stated practices from your suggestions. Don’t invent policies, approval limits, or standards; mark gaps as “Needs clarification.” Finish by asking what the guide oversimplifies or gets wrong.
Test It With the Person Doing the Work
Start with one recurring responsibility, something like reviewing a weekly project update rather than “running operations.” When you catch yourself saying “it depends,” explain what it depends on.
Then read the guide critically. AI can make a preference sound like a requirement or turn a situational call into a universal rule. Correct those, and make sure the decision boundaries match the authority you’ve actually agreed to delegate.
Finally, walk through a fresh example with your colleague. Ask what they notice, what they’d do, and where the guide leaves them unsure. Their questions will surface what you forgot to explain, and they may propose an approach you wouldn’t have chosen that works just as well. Leave room for that. The aim is not a copy of your judgment but a shared starting point from which they build their own, and the work that kept returning to your desk can finally stay where you handed it.
