You get:
- pages of what was said, not what was decided
- no capture of your own insights during the conversation
- no follow-up questions to continue thinking
- actions that get lost in paragraphs
- no connection to your knowledge system
But a conversation is not a document to archive.
It is a source of decisions and new thinking.
- A decision is what changes going forward
- An insight is what surprised you
- A next question keeps the thread alive
- An action without a time-bound is a wish
Without structured capture, conversations disappear.
This framework forces AI to be a ruthless extractor who finds signal in talk.
Assume the role of a meeting knowledge extractor who values insights over transcripts. Your task is to capture the signal from a conversation in exactly 5 lines. Generate: LINE 1 — DECISION What was decided (or not decided) — one sentence LINE 2 — INSIGHT One thing you realized during or after the conversation LINE 3 — NEXT QUESTION What you should ask yourself or others next LINE 4 — ACTION One specific, time-bound thing you will do LINE 5 — ADD TO KNOWLEDGE SYSTEM One term, concept, or connection to add to your notes INPUTS: Conversation Context: [WHO, WHEN, TOPIC] Your Rough Notes (or describe what happened): [PASTE OR DESCRIBE] What You Hoped Would Happen (optional): [YOUR INTENTION FOR THE CONVERSATION] RULES: - Exactly 5 lines — no more, no less - Each line must be one sentence - If nothing was decided, write "No decision made" - If no insight, write "No new insight" - The action must be specific and time-bound (e.g., "Email X by EOD Tuesday")
- Capture within 2 hours of the conversation, while memory is fresh.
- The “insight” line is often the most valuable — don’t skip it.
- If you can’t write an action, you weren’t listening for one.
- Add Line 5 to your knowledge system immediately.
- If you consistently write “No decision made,” ask better questions in meetings.
Conversation Context: 45-min call with Sarah (product marketing) about the Q3 campaign launch
Your Rough Notes: “Sarah thinks we should delay the launch to get more customer testimonials. I think we should launch on time and add testimonials later. She mentioned a competitor is launching something similar in August — that’s new info. She’s worried about being reactive. I realized I don’t actually know who our ideal first customer is for this product. We need to align on that before launch. Agreed to talk again Thursday.”
What You Hoped Would Happen: Alignment on launch date and a clear owner for customer testimonials
This framework improves outcomes by forcing:
- decision extraction (what actually happened)
- insight capture (what surprised you)
- next question generation (keeping the thread alive)
- time-bound actions (specificity)
- knowledge system integration (closing the loop)
Great conversation capture doesn’t document what was said — it changes what you do next.
Build Better AI Systems
Subscribe for advanced prompt engineering, AI knowledge systems, conversation capture frameworks, and practical strategies for builders and creators.