AI at work

Ask ChatGPT Questions First: Turn a Vague Task into a Usable Brief

You do not need a perfect prompt before asking for help. But an unlimited interview can consume the time you hoped to save. Ask questions that change the work, then define when to stop and draft.

The handover situation and dialogue below are fictional examples, not a recorded AI session or a policy recommendation for every organization.

Preserve the facts you have already supplied

I need a handover guide for one person taking over the inquiry desk next month.
The handover lasts two weeks. The new colleague has completed basic product training.
Reply procedures exist, but staff disagree about when an inquiry should go to a manager.

“Who is the guide for?” and “Have they had training?” have already been answered. The missing information is what decisions the guide must support and which rules are actually approved.

Set a short, decision-focused interview

Do not write the handover guide yet.
Ask the most important missing questions one at a time, at most three in total.
Ask about matters that materially change the work before asking about tone or layout.
Do not repeat information already supplied.
Briefly explain why each question is needed.
Offer choices where useful, while allowing a free-text answer.
If I answer “not decided,” do not convert a suggested default into a fact.
After my answers, separate confirmed facts, unresolved matters, and the draft that can be made now.

Three is a limit for this example, not a universal optimum. The aim is to avoid refining formatting while the substantive task remains unknown.

Different answers lead to different documents

Question Answer in this example Consequence
Will the guide be used before work, during a difficult inquiry, or for reporting afterwards? During an inquiry, when deciding whether to escalate. Build a short decision aid, not a long training course.
Which escalation rules are approved? Personal-data requests and compensation questions go to a manager. The rest is not standardized. Include those two rules; do not invent a complete policy.
Who reviews and approves additional rules? The inquiry-desk lead, at next week's meeting. Prepare a review draft with explicit open questions.

If the first answer had been “before starting the role,” the same source material might instead support an induction checklist. That is why use comes before formatting.

“Not decided” does not always mean “do nothing”

Situation Confirmed handling Still to check
Personal-data request Refer to a manager Contact method and urgent route
Compensation question Refer to a manager What to do if the approver is unavailable
Other uncertain inquiries Rules not yet standardized Which scenarios the lead must define

This is a draft of the fictional organization's stated rules, not a generally applicable escalation policy. The unanswered questions are part of its usefulness: they show precisely what the lead must decide.

Use confirmed facts as facts.
Leave matters requiring approval marked as awaiting confirmation.
You may propose headings and order, but label them as editorial suggestions.
Do not propose an approved compensation limit or other business rule unless it is in the supplied material.

Redirect an unhelpful interview

If the AI asks about headcount again, reply: “The original brief confirms one new colleague. Move that to confirmed facts and continue with the remaining question.” If it asks about tone too early, say: “Tone can wait. First clarify the decision the guide must support.”

The stopping condition is not that every conceivable question has been answered. It is that you can distinguish the document you can draft now from the decisions someone else still owns.

Use Prompt Ready to state the uncertainty clearly

Include the task, known background and intended output in Prompt Ready. When the material is incomplete, add “ask up to three important questions first” and “keep unapproved rules unresolved.” Review the generated request before copying it to your usual AI. Prompt Ready helps frame the request; neither it nor the AI should become the authority for the organization's missing rules.