Business
Chaos → decisions — action items with owners, open questions, and a narrative recap you can paste into Notion or email without inventing attendance.
If it is not written, it did not happen.
Paste rough notes, auto-transcripts, or Otter exports. Pick a meeting type so tone matches standup vs exec vs sales. The output foregrounds decisions and action items with explicit owners (when notes name them) and honest TBDs (when they do not). The model is instructed not to fabricate commitments, attendance, or numbers that do not appear in the raw notes — it surfaces ambiguity instead of papering over it. The result is a recap your team can paste into Linear, Asana, or email and act on without re-listening to the recording.
Six prep moves that beat re-watching the recording.
Each type shapes the output structure.
Status sweep
Round-the-room blockers and progress with emphasis on what changed since yesterday.
Scope + estimate
Story-grouped recap with capacity calls and parking-lot items for next planning cycle.
Options + chosen path
Decision log format with rationale, alternatives considered, and follow-up owners.
Career + work
Two-axis recap separating personal growth threads from immediate work commitments.
Pain + next step
BANT-shaped notes with named pain points, decision criteria, and committed follow-ups.
Crisp, decision-ready
Top-line summary with risks and asks separated for executive readability.
Recurring meetings where structure compounds.
Inventing owners is how trust dies in distributed teams.
The most damaging meeting-notes failure is not bad formatting — it is confidently assigning an action item to someone who never agreed to it. Generic AI summarizers love to fill in owners because the recap reads cleaner that way. This template inverts that incentive: when ownership is unclear in the raw notes, the output marks TBD loudly and surfaces it as a follow-up question. When dates are not stated, they stay blank instead of becoming "end of week." The recap reads slightly less polished than fabricated alternatives, but every commitment in it is real.
Habits that compound across recurring meetings.
It mirrors ownership language from your raw notes; if no owner is named, it writes TBD honestly. Blame is a human conversation, not a recap output.
It is explicitly instructed not to. If something appears that you did not write, fix the raw notes input and regenerate.
Use the AI Transcription tool first to clean audio into text, then paste the transcript into this tool for structured minutes.
Best in English today; other languages produce reasonable structure but ownership and date inference is weaker. Always verify.
Not directly — but the action items table is shaped for easy copy-paste. Each row maps cleanly to a ticket.
Default streaming text models work well for most meeting summaries. Use a deeper reasoning model for complex multi-decision strategy meetings.
Paste richer raw notes — three bullets of context per topic produce better one-line summaries than a wall of unstructured text.
Minutes only matter when linked to tickets.
This output is structured for the paste-into-PM-tool hop. Send the recap within an hour, assign the TBDs by the next meeting, and watch how much faster decisions stick when there is a written record of who agreed to what.