An archive becomes a memory the day you start interrogating it. Most people never do, because they assume the answer would take an hour of scrolling. With semantic search and chat over your own notes, these ten questions take a minute each—and several of them are worth asking regularly.
The ten questions
Each of these works because it asks about something your history actually witnessed: decisions, promises, opinions, and repetition.
- What did I decide about X—and when exactly?
- What have I promised, and to whom, that is still open?
- What is still owed to me that I have stopped tracking?
- What was I worried about six months ago—and did it happen?
- Which ideas keep coming back in different words?
- What did person N and I agree on last time we spoke?
- What did that repair, service, or subscription actually cost?
- Why did I choose this option over the alternative?
- What did I postpone ‘until later’—is later now?
- How has my thinking about Z changed over the past year?
Judge answers by their evidence
A good answer to any of these shows its sources: the notes it drew from, with dates and excerpts. If an answer arrives without evidence, treat it as a hypothesis and open the underlying notes before acting.
Chronology is half the answer
Questions about decisions and opinions are really questions about sequence. Insist on knowing which note is most recent; an old enthusiasm quoted without its later reversal is worse than no answer.
Make one of them a habit
Pick the question that hurts most when unanswered—for most people it is the one about open promises—and ask it every week. One recurring question turns an archive into a practice.
How to phrase a question your archive can answer
The same information can be requested in two ways, and only one of them works. ‘Am I spending too long on this client?’ asks for a judgment your notes never recorded. ‘What did I write about this client in the last two months?’ asks for material that exists.
Three habits do most of the work. Name at least one anchor—a person, a project, a rough period—so retrieval has something to hold onto. Ask about events rather than qualities: what was said, decided, quoted, promised. And when the first answer is too broad, add exactly one filter rather than rewriting the question, because a date range or a name usually collapses the candidate set immediately.
Three questions that will disappoint you
Not every question is answerable, and knowing which ones fail is part of using the archive well. Counting questions—‘how many times did I mention this?’—produce estimates, not totals. Retrieval returns the most relevant passages, not an exhaustive set, so any number it gives you is a floor rather than a count. Verify by hand if the number matters.
Negative questions are the second class: ‘which suppliers did I not contact?’ requires reasoning about absence, and an archive can only speak about what is present. Rephrase as a positive question—‘which suppliers did I contact?’—and do the subtraction yourself.
Questions about things you never captured are the third, and they are the most dangerous, because a fluent answer assembled from adjacent notes looks exactly like a correct one. This is why every answer must show its sources: seeing that a conclusion rests on two notes from January tells you what the system does not know as clearly as what it does.
Your history already contains the answers. The skill is asking specific questions and refusing any answer that cannot show where it came from.