You made a decision six months ago. It seemed obvious at the time. You moved on. Now it is causing problems. Someone asks why you did it that way. You have no idea. The reasoning felt clear then. It is gone now.
This happens to every knowledge worker. Decisions accumulate silently. The logic behind them evaporates. Six months later you are debugging choices with no documentation of why they were made.
Issue #8 is about fixing that — with a decision log that takes 3 minutes per entry and is actually useful.
Why Most Decision Logs Fail
You have probably tried keeping a decision log before. Most people have. Most people stopped.
The reason: the logs captured what was decided, not why. “Chose vendor X over vendor Y.” Great. Why? Unknown. What alternatives were considered? Unknown. What would trigger revisiting this? Unknown.
A decision log that only records outcomes is just a changelog. It does not help you learn.
The version that works captures three things: the reasoning at the time, the alternatives you rejected, and the conditions that would change your answer. Your AI helps you do all three in minutes.
The Decision Entry (5 minutes)
Run this immediately after any significant decision — hiring, architecture, strategy, vendors, pricing. Capture the reasoning while it is still fresh.
I just made this decision: [state the decision in one sentence] Help me document it properly. Ask me: 1. What specific problem was I solving? 2. What alternatives did I consider and why did I reject them? 3. What assumptions am I making that could turn out to be wrong? 4. What would have to change for me to reverse this decision? 5. How will I know in 30/90 days if this was the right call? Then format my answers as a structured decision log entry.
Why this works: The AI asks you the questions you would skip if you were writing this alone. Most people will document the decision but not the rejection criteria, not the assumptions, not the reversibility conditions. Those are the parts that matter.
The difference: Five minutes now. Saves two hours of confusion later.
The Pre-Mortem (before you decide)
This one is used BEFORE a major decision, not after. It is the fastest way to surface assumptions you did not know you were making.
I am about to make this decision: [state the decision] Run a pre-mortem. Assume it is 12 months from now and this turned out to be a serious mistake. What went wrong? Give me 5 specific, plausible failure modes — not generic risks, but the actual ways this specific decision could fail given what I have told you about the situation. Then: for each failure mode, what is one thing I could do NOW to either prevent it or reduce its impact?
Why this works: Most people skip this step because it feels pessimistic. It is the opposite — the failure modes it surfaces are usually obvious in hindsight. The point is to make them obvious now, when you can still act on them.
The test: If a failure mode surprises you, the pre-mortem just paid for itself.
The Weekly Review (10 minutes)
Once a week, review the decisions you logged. This is where the compounding happens — a feedback loop between your past self’s reasoning and your present self’s information.
Here are the decisions I logged this week: [paste 2-5 entries] For each one: 1. Has anything changed that affects the underlying assumptions? 2. Is there any new information that should update my confidence in this decision? 3. Are there any patterns across these decisions — am I consistently optimizing for one thing at the expense of another? Be direct. If something looks like a mistake in progress, say so.
Why this works: After three months of weekly reviews, you have something most people never build: a documented model of how you make decisions, where your blind spots are, and which assumptions tend to be wrong.
The compound: Each review makes the next decision slightly better. That is how decision quality improves.
The 3-Minute Version
If the full system feels like too much, this is the minimum viable version:
I decided to [X]. My reasoning was [Y]. The main risk is [Z]. I will revisit this on [date].
Three sentences. One minute. Paste it into a running document. That is enough to be dramatically better than nothing. Add structure when you find yourself needing it.
The Learning Flywheel
After 90 days of keeping a decision log, something changes. You start making decisions differently — not because you are smarter, but because you are clearer about what you are actually deciding.
You stop confusing reversible decisions (try it, see what happens) with irreversible ones (requires a pre-mortem). You start noticing when you are making the same kind of mistake repeatedly. You build a library of your own patterns.
Your AI becomes the analyst that surfaces those patterns across entries, flags when you are repeating a past mistake, and pushes back when your stated reasoning does not match what you did last time.
5 min per decision now Saves 2 hours of confusion later
A documented model of how you make decisions — and where your blind spots are.
Try It Today
Think of a decision you made this week. Any decision. Run Prompt 1 on it right now. It takes 5 minutes. Pay attention to what the AI asks you that you would not have written down on your own.
Then reply to this email and tell me what you captured. I read every response.
The Reasoning Audit: How to use AI to pressure-test your thinking before you act on it, not after — and the prompt that catches the assumptions you did not know you were making.