10 min · Intermediate
Decisions, Notes, and a Learning Queue
Draft a decision memo, theme a messy dump, and keep learning to a 30-minute block tied to real work.
Learning objectives
- Draft a decision memo with options, criteria, a recommendation, risks, and a review date.
- Turn a brain dump into themes and open questions without invented answers.
- Separate a quantity brainstorm from the one option you might actually try.
01
Why this matters
Decisions get muddy when notes, ideas, and commitments sit in one blob. A memo, a themed dump, a short study block, and a three-pass brainstorm keep those jobs apart. You leave with a draft and a choice, not with a plan that quietly signed itself.
02
Concept
A decision memo is a short record, not a vibe. Name the decision in one line. List real options, including wait. Set criteria such as cost, time, risk, and whether you can undo it. Ask for one recommendation and a reason the other options lost. Add risks and a review date, the day you will look again. Gemini may draft the memo. You still choose when money, people, or safety are involved.
Start from the messy dump, not from a fake template. Ask for themes, then open questions under each theme. An open question is something the notes do not answer. Tell the model not to fill those holes with guesses. Park items that are interesting but not part of this week's outcomes. You want a map you can scan, with holes left visible. A tidy summary that secretly answers the holes has done harm.
Tie learning to work you already have. Choose one skill that would help a task you already have. Ask for a single 30-minute weekly block with a recall step and a tiny application. Recall means you try to remember before you reread. Reject any plan that needs four hours or a new personality. You pick a source you trust and you do the half hour. If you miss a week, shrink it.
Use three passes, and do not mix them. First ask for quantity, including odd options, with no critique. Second, add constraints you really face, such as time, budget, and rules. Third, pick one option to try and park the rest. Say in the prompt that idea generation is not a commitment. A clever line must not become a promise to a customer. A founder might leave with one small experiment.
03
Real-world example
Sunday hours, a dump, and thirty minutes
Lina runs a neighborhood bakery and must decide whether to add Sunday hours. She asks for a memo with the decision line, three options including stay closed, criteria of staff fatigue, extra cost, and possible revenue, one recommendation, risks, and a review date four weeks out. Gemini recommends a four-hour trial. Lina rejects it because her baker already works six days, a fact she adds in the next prompt. Separately she dumps messy notes from a course on bread pricing and asks for themes and open questions, with a ban on invented numbers. For learning she asks for one 30-minute Friday block on costing a single loaf, with three recall questions and one worked example using her own prices. She does not adopt a four-hour study plan.
04
Gemini demonstration
A realistic exchange. Open Gemini beside this page and try a version with your own facts.
You
Decision: whether our team switches the Friday standup from 30 minutes to 15. Options I will consider: keep 30, cut to 15, or alternate. Criteria: meeting load, whether blockers still surface, and whether remote staff can hear updates. Do not add a fourth option that needs new software. Draft a memo with a recommendation, risks, and a review date. I have not decided.
Gemini
Decision to draft, not a decision made: whether Friday standup stays 30 minutes, moves to 15, or alternates. Options are only the three you named. I am not adding a tool. Criteria you gave: meeting load, blockers still surfacing, and remote staff hearing updates. Recommendation as a proposal: alternate for four Fridays, because a permanent cut risks silent blockers, and staying at 30 keeps the load you already dislike. Risks: people skip the long weeks, and remote staff talk over each other in the short weeks. Review date: four Fridays from the start, which you must set, because I do not know your calendar. Unknown: how many people dial in, and whether blockers are already tracked in a ticket list. You commit. I only drafted the comparison.
You
Remote staff already share blockers in the ticket list before Friday. That fact removes part of your risk. Also do not turn this into a commitment email. Separate a brainstorm of ways to run the 15-minute version, quantity first, then apply the constraint that we add no software, then pick one to try. Mark it as not committed.
Gemini
Updated proposal, still not a commitment. Because blockers already live in the ticket list, the risk of a shorter standup is smaller than I said. I withdraw the claim that a 15-minute slot will hide blockers. That was my guess, not your fact. Brainstorm, pass one, no critique: start with a one-line ticket readout, use a timer, let one person speak, collect questions in chat, rotate the facilitator, end with a single ask. Pass two, your constraint: no new software, so chat means the tool you already have, and the timer can be a phone. Pass three, one to try, not a promise: a timer plus a one-line ticket readout for four Fridays. Park the rest. I will not email anyone. You decide whether to trial it.
05
Prompt breakdown
The prompt
Decision memo. Decision line, options including wait, criteria, one recommendation, why the others lost, risks, and a review date I must confirm. Do not commit. Then brainstorm in three passes: quantity, then my constraints, then one option to try. Idea generation is not a promise.
- Including wait stops the memo from treating action as the only serious choice.
- One recommendation is easier to accept or reject than a cloud of maybes.
- The review date makes the choice temporary, which you still have to put on a real day.
- Three passes keep a long list from turning into a customer promise by accident.
06
Common mistakes
Usually
Decide for me and email the team that we are doing it.
Better
Draft the memo and stop. I will choose. Do not write a sendable announcement.
Usually
Clean up my notes and fill any gaps so the page looks finished.
Better
Theme the dump and list open questions. If the notes do not answer it, do not guess.
Usually
Build me a four-hour daily study routine and a new personality by Friday.
Better
Give me one 30-minute block this week tied to a task I already do, with a recall step.
07
Practice exercise
Memo, map, and a half hour
Done when: You can point to a recommendation you have not accepted, a list of open questions, and a 30-minute block that fits a real week.
08
Challenge
Refuse one recommendation on purpose
Pick a real decision you have not made, small enough to finish this week, such as a schedule change or a purchase under a limit you set. Ask for a memo with options, criteria, a recommendation, risks, and a review date. Reject any option the model added that you cannot actually carry out. Separately, dump a page of messy notes and ask for themes and open questions only. Then ask for a 30-minute study block tied to one skill that decision needs. Do not start a four-hour plan. Write down which recommendation you refused, and the fact that made you refuse it.
Hint: A four-hour plan is a signal to shrink the block, not a dare to try harder.
09
Key takeaways
- A decision memo recommends. It does not commit, especially where money or safety is involved.
- Open questions should stay open when the notes do not answer them.
- A learning queue is one 30-minute block tied to your work, not a fantasy routine.
- Brainstorm quantity, then constraints, then one trial. Generation is not a promise.
10