Build a Decision Record You Can Learn From

September 15, 2026 · By Michel Valenzuela, Founder of Daryl

Three months ago, you made a decision that seemed sensible.

Now the outcome is disappointing, and you are trying to understand why.

What did you know at the time? Which alternatives did you consider? Was the problem foreseeable, or did something change afterward?

Without a record, you have to reconstruct the reasoning from memory.

A short decision record gives you something more concrete to examine. It preserves the circumstances behind the choice so you can learn from what happened next.

AI can help you prepare that record while the discussion is still fresh.

Capture the decision while you understand it

You do not need a long report.

For an ordinary project decision, a few clear paragraphs may be enough. Record what you chose, why you chose it, and what remained uncertain.

Imagine accepting a client engagement with a limited initial scope.

Your record might say:

“We accepted the first phase because it fits our available capacity and allows us to assess the working relationship before committing to the full project. The client confirmed two deliverables and one review round. Further work will require a separate agreement.”

That gives your future self a useful starting point.

It also helps anyone involved understand the boundaries of the commitment.

Preserve the information available at the time

The record should reflect what you actually knew.

A confirmed delivery date belongs in a different category from an estimate. A client’s written agreement carries a different weight from your impression of a conversation.

If you used AI to examine the decision, ask:

“Summarise the facts, estimates, and assumptions we used. Keep those categories separate and identify anything that still needs confirmation.”

Then check the summary against the source material.

The assistant may compress an uncertainty into a statement that sounds settled. Correct that before preserving the record.

Its usefulness depends on faithfully representing the basis of the decision.

Explain why the alternative lost

The option you rejected can teach you something later.

Perhaps you declined a larger engagement because it would delay an existing commitment. Perhaps you chose a slower approach because you could support it more reliably.

Record the main reason.

You do not need to catalogue every possibility discussed. Preserve the alternatives that were credible and the trade-offs that determined the choice.

This helps you revisit the decision without reopening the entire conversation.

If the constraint changes, you can see whether the rejected option deserves another look.

State what you expect to happen

A decision often contains an expectation.

You believe the smaller release will produce useful feedback. You expect a revised process to reduce repeated work. You think a limited engagement will reveal whether a longer partnership is worthwhile.

Write that expectation plainly.

For example:

“After the first phase, we expect to understand the client’s review process well enough to estimate the next phase more accurately.”

That gives the later review a question to answer.

Avoid making the expectation more precise than the evidence supports. An unsupported number can create an appearance of rigour without improving the record.

Define when to review it

Some decisions need a scheduled review. Others need reconsideration only if a particular condition changes.

A trial engagement might be reviewed at the end of the first phase. A delivery plan might need reassessment if a dependency slips.

State the trigger:

“Review before agreeing to phase two, or earlier if the scope exceeds the confirmed deliverables.”

That creates a clear connection between new information and the original choice.

It also prevents a record from becoming a document nobody returns to.

Keep the record proportionate

A small, reversible decision should not require extensive administration.

For a larger commitment, more detail may be useful, particularly when several people are involved or the consequences are difficult to reverse.

A practical structure is:

The format should make the decision easier to understand and review. If maintaining it becomes more work than the decision warrants, simplify it.

Use AI to organise the record

After a useful conversation, you can ask:

“Create a concise decision record from this discussion. Include the chosen direction, supporting reasons, alternatives, unresolved assumptions, and review point. Do not fill gaps with invented details.”

Read the result before treating it as authoritative.

If the discussion never established a review date, leave that open until you choose one. If responsibility is unclear, resolve it directly.

A well-written summary still needs your confirmation that it represents the actual decision.

Review both the reasoning and the outcome

When you return to the record, ask what happened and why.

Did the assumptions hold? Did you follow through on the agreed action? Did new circumstances affect the result? Was there information you could reasonably have obtained before committing?

A disappointing outcome does not automatically establish that the original reasoning was poor. A successful outcome does not prove that every assumption was sound.

Examine both.

The aim is to identify what you can improve: a question you should ask earlier, an estimate that needs more care, or a condition that deserves explicit confirmation.

Where Daryl fits

Daryl is a personal AI for Mac, currently in development, designed around local conversation and support for decisions.

I want the experience to help people understand the basis of a choice and carry useful context forward under their control. Local memory and continuity are part of the development roadmap.

This article describes a practice you can use with AI. It does not announce a released decision-journal feature or automatic record keeping in Daryl.

For now, a short record in a document or notes application can serve the purpose.

The important thing is preserving enough of the reasoning to make the next review worthwhile.

A decision shapes what happens next. A record helps you understand what that experience can teach you.