The task
You sat through a 60-minute meeting and now you have a 14,000-character transcript. The default failure is a “summary” that no one re-reads: ten bullets, no owners, no dates, no path to follow-up. The goal is the opposite — strip the discussion, keep the outcomes, attach owners and deadlines to every action item, and surface open questions so they reach the next meeting instead of disappearing.
When AI helps — and when it does not
AI is excellent at structuring an unstructured transcript into decisions / actions / questions, and at tagging owners when the transcript names them. It is poor at understanding implied owners (“we’ll handle it”), commitments without dates, and chemistry / political context that decides whose action is actually load-bearing. Mark anything ambiguous as “needs confirmation”; do not let AI guess.
Workflow
- Record / transcribe: Otter, Fireflies, Granola, Notta, Apple Voice Memos + Whisper. Verify the transcript has speaker labels; without them, action attribution falls apart.
- AI structuring: feed the transcript to ChatGPT / Claude with the prompt below.
- Structured output: decisions, actions with owner + deadline, open questions, next-meeting agenda.
- Push to your task system: Linear, Jira, Asana, Notion. Actions that stay in the notes die.
What to feed the AI
- The transcript with speaker labels (Speaker 1 / Speaker 2 / …)
- Meeting topic + purpose
- Who attended (role + name) so AI can resolve speakers
- Decisions that were definitely made before the meeting (so AI does not double-count them)
- The team’s working convention — what counts as an action (“only items with explicit verbs”), what counts as a decision (passive consensus or explicit vote)
- Where the output will live (Notion page, Slack thread, ticket descriptions)
Copy-ready prompt
You are an experienced meeting secretary at a tech company.
Meeting topic and purpose: <line>
Attendees and roles: <list>
Pre-existing decisions (do not re-list): <list>
Action convention: <only explicit verbs / include implied>
Transcript:
"""
<paste>
"""
Return:
1. Meeting topic and purpose (≤1 sentence)
2. Key decisions — each tagged with decision-maker and when it takes effect
3. Action items — each must include: action description, owner, deadline. If owner or deadline is missing in the transcript, mark "needs confirmation."
4. Open / unresolved questions — with the person who can answer
5. Suggested agenda for the next meeting
6. A "watch out" — anything in the transcript that sounded like a commitment but was hedged
Do not restate the discussion. Outcomes only. If an action was discussed but never agreed, list it under open questions, not actions.A second pass for long meetings: “Now identify the 3 highest-leverage action items — the ones that, if dropped, would block the most other work.”
Recommended output structure
A short header, decisions in a list with decision-maker per line, actions in a table (action / owner / due / status), open questions in a list with answerer, next-meeting agenda as 3 bullets. The whole note should fit on one screen.
How to check the output is usable
- Every action has owner + deadline, or is explicitly marked “needs confirmation”
- Decisions are not duplicated as actions
- The next-meeting agenda is short and specific
- “Watch out” surfaces hedged commitments — they are the most common dropped balls
- A person who missed the meeting can read this and know what to do
Common mistakes
- Letting AI summarise the discussion — nobody re-reads summaries
- Inventing owners or deadlines — mark as “needs confirmation” instead
- Mixing decisions and actions in one list — they have different follow-up paths
- Not pushing to the task system — notes that live in a doc do not move
- Multi-language meetings without language tag — AI sometimes drops the non-dominant language. Tell it explicitly.
Practical depth notes
For How to Use AI for Meeting Notes: From Recording to Executable Action Items, the difference between a usable AI result and a generic one is the input packet. Give the model the audience, the current draft or raw material, the desired format, the decision you need to make, and two examples of what good and bad output look like. Ask it to preserve facts first, then improve structure or wording second.
After the first response, do a separate review pass. Look for missing constraints, invented details, weak calls to action, and language that sounds plausible but does not match the real situation. The best final output should be easy to use immediately: clear owner, clear next step, and no hidden assumption that someone else has to untangle.
FAQ
- What about multilingual meetings (zh + en code-switch)? Tell AI to preserve the original language for quotes; translate only summaries.
- How long after the meeting? Within 24 hours, ideally same-day. Memory is the verification layer.
- Should AI write the next-meeting invitation? Yes, with agenda pre-filled from open questions.
Related
- Study notes cleanup — same approach for lectures
- Project plan draft — when the meeting produces a project
- Feature prioritisation — when meetings sort the backlog
- PRD draft — meetings → PRD draft
- Meeting notes prompts — additional prompt variants
- AI meeting summary tutorial — full workflow
- AI status update tutorial — turn notes into status updates
Tags: #AI writing #Prompt