Stop retyping the same questions: build a library of AI master prompts

Most people using AI at work retype the same question every time they open the chat window. Slightly different words, slightly different output, twenty minutes gone. Then they do it again on Thursday.

In my restaurant we would never let a team member improvise a service step every shift. We write it down once, refine it, and run it the same way every time. Your best AI questions deserve the same treatment. Build the prompt once, save it, run it forever. Think of it as a standard operating procedure for your own judgment.

I keep a library of these. Here is what surprised me after a year of running it: the prompts that changed how I lead are not the ones that save me time.

The two shelves

My library has two shelves.

The first shelf is efficiency. Drafting SOPs from a description of how we actually do something. Turning a brain dump into a meeting agenda. Drafting a professional vendor email that says what I need without me stewing over it for half an hour. These are real. They gave me back hours every week, and I am not giving them back.

The second shelf is the one nobody talks about. Those prompts point the AI at my people instead of my tasks.

“Who on my team deserves recognition that has not received it?”

“Who is least likely to ask for help but most likely to need it?”

“Who might have left work today feeling invisible?”

Same AI. Same chat window. Completely different job.

What the second shelf did

I ran that first question against my own team not long ago. It gave me a name I did not expect: a kitchen team member who had been coming in early, again and again, to cover for a teammate who was out for health reasons. Never mentioned it. Never asked for credit. Invisible in every report I normally read, except one, which had flagged her the week before as an overtime risk.

One question made her a cost. The other made her the person holding the kitchen together. The AI did not change. The question did.

The efficiency shelf buys the time. The significance shelf tells you what the time is for.

How to build yours

Do not overthink the mechanics. Pick a task or a question you return to at least weekly. Write the prompt longer than feels natural: who the AI should be, what you want, what good looks like, what to avoid. Run it three times and refine between runs. Then save it in one place you will actually check. A doc, a note, a text file. The tool does not matter. The habit does.

The part most people skip: put at least one people question in the library. You will run the SOP prompt when work demands it. Nothing will ever demand that you ask who feels invisible. That one only runs if you decided in advance to run it, which is exactly why it belongs in the library.

At the end of my keynotes I ask everyone to take out their phone and send one sentence of encouragement to someone who had an impact on them. Grown adults tear up over a text message, every single time. That is thirty seconds of intentionality against years of it going unsaid. A saved prompt is the same discipline pointed at your calendar: care, scheduled, so it actually happens.

Steal mine

I published my whole people library, numbered, at jonharmeling.com/prompts. Take whatever is useful. Build the efficiency shelf too, because the hours are real.

Then spend the hours on the second shelf. That is the whole system.