Two useful questions
A better prompt can make an answer clearer, more specific, or easier to use. That is valuable when the problem is an unclear instruction.
But a well-written instruction can still leave out someone who matters. A proposal may explain your service beautifully while saying little about the customer’s implementation burden.
Ask both questions: Have I explained the task clearly? And whose perspective have I left out?
The difference in practice
| Improve the prompt | Add a perspective review | |
|---|---|---|
| Start with | What should AI produce? | Who needs to act on this work? |
| Look for | Clarity, structure, and useful detail | Missing assumptions, questions, and consequences |
| Next step | Refine the output | Sort the findings and check what matters |
These approaches can work together. Use a clear Ask, then review the answer before treating it as ready.
One proposal, two reviews
Situation: You are proposing a new customer onboarding process.
A prompt review asks for a shorter, clearer proposal. The result is easier to read.
A perspective review considers the support lead who must run the process. Who handles exceptions? How much training does the team need? What happens during a busy week?
The second review suggests questions the first may leave unanswered. Those are hypotheses to investigate, not proof that the support lead objects.
The next move: Check the staffing and training assumptions with the support lead. Revise the proposal around what you learn.
Try it on your next draft
- Name the decision or action the draft needs to support.
- Choose one person closest to that outcome.
- Describe their situation and review the draft from that seat.
- Sort findings into Use, Validate, Explore, or Reject.
- Check important assumptions before making the next change.
A stronger sentence is useful. So is noticing the question the sentence never answered.
What changes when you start with perspective?
Prompting shapes what you ask AI to produce. The Empty Chair method adds a review: whose situation does this work need to account for? Start with the person closest to the next outcome, consider the draft from their perspective, and verify important assumptions before you act.