Make a Decision Between Two Options You Can Review Later
Decisions stall less often from missing information than from unstated assumptions and no one arguing against you. This workflow supplies both.
How to do it
Step 1: Lay both options out in one table
Use “Choosing Between Two Options” to propose four comparison dimensions and explain why those four, then fill each cell: cost, latest decision date, reversibility, and what you lose if it fails. Build the table first and resist asking for a recommendation. Put your own constraints in the brackets; different weights flip the conclusion. Leave a cell blank rather than filling it with nice wording.
Step 2: Surface the premises with “Unpack My Assumptions”
Ask it to sort the assumptions behind each option into three buckets: fact, inference, preference. Then ask of each one: if this is false, does the option still hold? You will usually find that the real disagreement is one or two lines, and the rest is packaging.
Step 3: Aim “Red-Team My Judgment” only at your favorite
Use “Red-Team My Judgment” and aim it only at the option you prefer. Have it play the opponent, list the three most likely ways you are wrong, and rank them by probability times loss. People quietly collect reasons for what they already want, so this is the one step that works against your preference. Afterward decide whether to change the plan or accept the risk knowingly.
Step 4: Set the stopping rules with “Set Exit Conditions Up Front”
Use “Set Exit Conditions Up Front” to decide the stopping rules before you start: three to five observable signals, a check date and threshold for each, and who makes the call. The key distinction is “temporarily rough” versus “wrong direction” — the first gets another chance, the second gets cut. Without this, the more you invest, the harder stopping becomes.
Step 5: Write the contingency with “What If I'm Wrong”
Use “What If I'm Wrong” to write the contingency: three most likely ways you are wrong, the earliest signal of each, what you do in the first 24 hours and the first week, and which fallbacks to prepare now. This is not pessimism; it turns “what if” into actions. Store this with the comparison table and review both in three months.
Prompts used
FAQ
Can I trust the advice an AI gives me?
It can put your assumptions and trade-offs on the table, but it does not know your budget, your team, or your risk tolerance. Use it as a table generator; the trade-off and the signature are still yours.
What if the two options look equally good?
That usually means the difference lives in the assumptions, not the options. Go get one piece of information that tests the key assumption, or pick the more reversible option and run it for a short stretch.
How do I make a decision reviewable later?
Write three things on the day: the assumptions it rests on, the signals you expect, and what would change your mind. Without them, the post-mortem is an argument about impressions.