I sit in four video calls most days, and half of them end with nobody remembering who owns what. These three prompts turn raw transcripts into clean decisions and action items so the call actually moves the project forward.
[Claude]
Given this meeting transcript, extract every decision made and every task assigned. For each task output a table with columns: owner, deadline mentioned, exact verbatim action, and the decision it traces back to. Then flag any task with no owner or no deadline as UNOWNED and list those separately. Do not summarize the conversation; only deliver the table and the unowned list.
[ChatGPT]
Act as a meeting minute editor. Here is a transcript with three speakers. Rebuild it as a structured handoff doc with: (1) decisions in bold, (2) owners with due dates, (3) open questions blocking next steps. Keep every owner name and date exactly as spoken, no guessing. End with a 5-line summary a new teammate could read in 60 seconds.
[Cursor]
Write a Python script that parses this transcript text and splits it into segments by speaker turns. Then tag each segment as decision, task, question, or filler based on keyword rules you define, and print a markdown report grouping all tasks with their nearest preceding decision. Save the report to meeting_notes.md. Handle missing speaker labels gracefully by defaulting to Unidentified.
More packs in the archive at aitoolalliance.com/tag/prompt-pack.