Your next good AI idea may need the exact person who is finishing your last good decision.
Picture a hypothetical growth team. You have promised to clean up the company’s service descriptions so customers can understand what is included before they agree to a job. Your editor is working through them. Then you see a useful way to have AI draft the customer guides that explain how to prepare for a visit.
The guides are real work. Someone writes them already. You can picture why help would be useful.
“Could you take a look at this while you’re doing the website?”
Same editor. Same afternoon. You have just offered the same stretch of time to two different jobs.
Here’s the thing. The AI idea does not have to be foolish for that to be a bad decision. It can be sensible, practical, and worth pursuing. It can still be the wrong thing to put in front of that person now.
The most dangerous AI ideas are not bad ideas. They are interesting ideas that never earned the right to become operating priorities.
Useful still has to get in line
If you lead a department, part of your job is spotting better ways to work. You notice possibilities the team is too busy to investigate. You want them to know that useful suggestions will get a hearing.
That is good leadership right up until every hearing becomes an assignment.
“Let’s just explore it.”
Explore sounds light. Somebody still has to gather the material, understand the idea, and decide whether it is worth another conversation. In our example, that somebody is the editor who knows the company’s services well enough to check what AI writes.
Her time is what makes the exploration useful. It is also what you already committed to the service descriptions.
You are not asking for much. Let me tighten that. You may be asking for something small from the one person whose afternoon is already spoken for.
Right?
The question is bigger than whether AI could help write a guide. You need to decide whether investigating that help deserves a place ahead of work you have already chosen.
That is where The Prep List belongs. It helps you choose which recurring job to work on next. In a kitchen, having a good recipe does not mean you can add another dish during dinner service. Someone has to prepare it, cook it, and check it while the other orders keep coming.
The Prep List looks at whether work repeats in a familiar shape, how much of the week it consumes, whether you can describe a good result, and how easily you can recover from a mistake. Those questions help an idea make its case.
They do not create another editor.
Let the idea make a proper case
Before you argue about priority, give the idea a fair hearing. Five questions help. They borrow from the Prep List’s selection logic, with an extra question about the information AI would need. They are a way to examine a proposal, not a score that proves it will work.
Start with how often the work happens. For the proposed guides, are people regularly writing similar explanations, or does every guide require a completely different answer? Work that returns in a familiar shape gives you something to learn from and reuse. A task occurring every day does not, by itself, tell you how much time it consumes.
Then look at the pain. How much of the editor’s working week goes into those guides now? Where does the difficulty sit: finding the right information, writing the first draft, or checking it? If nobody knows, leave that answer open. A busy person saying “these take forever” gives you a reason to look. It does not give you a measured saving to put in the proposal.
Ask about judgment. Can someone describe a useful guide and recognize one when it is wrong? The editor might need to spot a promised service the company does not provide, an instruction a customer could misunderstand, or an important preparation step that is missing. Put a name beside that checking work. Then find out whether that person has time to do it.
Ask about the inputs. Are the approved service details and existing public instructions available for this use? “We can probably find that somewhere” leaves a job unfinished. An idea waiting on missing material or permission waits, however attractive the other answers look.
Finally, ask what happens when it is wrong. For this hypothetical, keep generated drafts away from customers until the editor has checked them. If you cannot hold the draft or catch a consequential mistake, you have more to resolve before this becomes a candidate.
Do not turn those five answers into a contest where enough good numbers can cancel out a missing reviewer or information you are not allowed to use. They cannot.
Now give our example the strongest fair chance. Suppose the work does repeat, takes enough time to deserve attention, has approved material available, and can stay in draft until a capable person reviews it.
That makes it worth considering. We have established a plausible project. We have not established that it belongs at the front of the line.
The question after all the good answers
Here’s the thing. This is where a curious leader can feel finished.
“It checks out. Let’s do it.”
You still owe the editor one answer: what should she stop doing to make room?
In our hypothetical, the current goal is to make the service descriptions clear before the next round of customer conversations. That work is already underway, with a promised review. The guide project offers a possible improvement to another recurring job. Nobody has yet shown that exploring it matters more than finishing the descriptions.
The useful decision is to keep the editor on the promised work and leave the guide idea waiting.
Yeah. Even after the idea passed all five questions.
Write down why it waits. After the service descriptions are reviewed, the growth lead and editor will revisit the guide project against the work next in line. The trigger is an observable event, not “when things calm down.” And a revisit promises a decision, not an automatic start.
That protects the idea as well as the current work. You can come back to the reason it was parked without recreating the whole conversation.
Sometimes the new idea should win
The old commitment does not win just because it got there first.
Suppose the conversation with the editor reveals that guide writing is repeatedly taking the time meant for the descriptions. Now the new proposal might address the reason the current priority keeps slipping. That is new information worth examining. It is still a possible explanation, not proof that AI will fix it.
Or suppose the company’s goal has changed, and the planned description cleanup no longer serves the most pressing need. Finishing it merely because it is underway can be its own waste.
You are allowed to change the order. The job is to make that change honestly.
Name the work that will wait. Name the person who relies on its promised completion. Have the person who owns that promise decide whether the trade is acceptable. Give the editor a clear answer about which job comes first.
“We are moving this ahead of the descriptions, and I am taking responsibility for changing that commitment.”
That is a different instruction from asking her to fit both in.
Does that make sense? The discipline is not loyalty to an old list. It is making the cost of a new yes visible while you can still decide whether to pay it.
Your Monday Move: make one priority decision
Take one AI project you want to pursue to the person who does the affected work. Give yourselves fifteen minutes to decide what you know and what deserves a place in the queue.
Write short answers to the five questions: how often the work repeats, how much working time it consumes, who can judge the result, whether the needed information is available and approved, and what happens if the result is wrong. Keep missing answers visible.
Then name the current commitment competing for the same person’s time. If there is unused capacity, name the actual space in the calendar. “They’ll find time” does not count.
Record one decision: admit the idea for planning, clarify a named missing fact, or park it. Include the reason, the work owner, and who can approve any changed commitment. For an idea that waits, write the event or evidence that would bring it back for a decision.
This is a priority decision. Buying a tool, changing access, or starting a trial still needs its own authority and preparation. Completing the conversation does not give you those permissions.
The visible result is one idea with a place, a reason, and a person responsible for the next decision. Your editor should be able to look at it and know which job she is doing first.
Keep being curious. Before your next good idea gets someone’s afternoon, name what moves out of its way.
Original framework. Distilled from client work.
