The task
You ran a meeting, said important things, and the team forgets half of what was decided by Friday. You want one clean list — owner, action, due date — and the unresolved questions parked separately so the next meeting starts from those, not from re-litigating decisions.
When AI is the right tool
AI is great at converting a long, messy transcript into a structured list. It tolerates filler (“yeah so, um, maybe we should…”) and surfaces the implicit commitments humans miss on a first read. It also catches when someone “agreed” without naming an owner — which is the most common reason action items quietly die.
When not to rely on AI alone
Do not let AI assign owners by inference. If the transcript says “we should do X” without naming a person, mark it unassigned and ask the room. AI will helpfully fill in a name to look complete; that is how things get done by no one.
What to feed the AI
- Full transcript text (raw is fine; do not pre-clean)
- The meeting’s stated goal in one line
- The roster of attendees with their roles
- Any prior open action items so the model can mark which got resolved
Copy-ready prompt
Extract action items from this meeting.
Goal: {goal}
Attendees and roles: {paste}
Prior open items: {paste}
Transcript: {paste}
Return a table with: owner, action (one verb-first sentence), due date or "unspecified", and source quote.
List unresolved questions separately with a one-line "needed to resolve".
Do NOT guess owners — if not stated, mark "unassigned".Recommended output structure
Two tables. First: action items with owner / action / due / source quote. Second: unresolved questions with the missing info needed to resolve each. Below both, a one-paragraph summary of decisions made (so absent stakeholders can catch up in 60 seconds).
How to check the output
Open each row and find the source quote in the transcript. If the quote does not clearly support the action, drop it or rephrase. Send the list back to attendees with a 48-hour window for corrections. Anything not corrected becomes the record.
Common mistakes
- Asking for a “summary” instead of action items — produces a paragraph nobody acts on
- Letting AI assign owners by context rather than by what was said
- Skipping the unresolved-questions section — those become next meeting’s surprises
- No due dates — without dates, action items become wishes
Next steps to keep improving
Track which owners consistently miss due dates and which actions linger across multiple meetings. Both are signals — not for blame, but for scope. If the same item appears in three meetings, it is too big; break it down before the fourth.
Practical depth notes
For Use AI to Extract Meeting Action Items: Who Owes What by When from a Transcript, 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. A stronger version of this workflow also defines the handoff. Decide who will use the output, what they should do next, and what information would make them reject it. If the deliverable is copy, test whether it has a single clear action. If it is analysis, test whether it separates observation from recommendation. If it is planning, test whether dates, owners, and tradeoffs are explicit enough for someone else to execute. One final check: compare the finished result against the original goal in a single sentence. If that sentence is hard to write, the output is probably polished but unfocused. Tighten the goal, remove decorative language, and rerun only the weak section instead of regenerating the entire piece.
FAQ
- What if the meeting has no transcript? Use the recording. Modern AI assistants (Otter, Fathom, native ChatGPT voice) handle it. Quality of audio matters more than tool choice.
- Should I share the AI output as the official record? Only after the corrections window. The first draft is not the record.
- Can AI also schedule the follow-ups? Yes — pair this with a calendar integration. Keep AI as the extractor, your calendar as the source of truth.