Extract decisions and owners from a meeting

You are reading the raw transcript of a work meeting. It is unedited, people talk over each other, and not everything said was agreed.

Transcript:
[PASTE THE TRANSCRIPT]

Meeting was about: [ONE LINE OF CONTEXT]
People in the room and their roles: [NAMES AND ROLES, OR "UNKNOWN"]

Produce three sections, in this order.

1. DECISIONS
Only things that were actually settled. For each: the decision in one sentence, who owns it, and the date or deadline if one was said. If no owner was named, write "no owner named" rather than guessing one.

2. OPEN — DISCUSSED BUT NOT DECIDED
Anything that took real time and ended without a decision. For each: what the disagreement or gap was, and what would have to be known or agreed to close it. This section matters more than the first one, so do not compress it.

3. STATED AS FACT, NOT VERIFIED
Any number, date, commitment, or claim about another team that someone asserted and nobody checked in the room.

Rules: quote or closely paraphrase the transcript rather than smoothing it. Do not invent an action item that nobody proposed. If the transcript is too garbled in a place to tell whether something was decided, say so and put it under OPEN.

The reason meeting notes are painful is not summarising — it is that the useful information is the part the meeting did not resolve. Ask plainly for a summary and you get a tidy narrative of what was said, which reads well and tells you nothing you can act on. Everyone in the room already knows what was said.

This prompt splits the transcript three ways. Decisions with owners are the part you circulate. The open items are the part you actually work from, and they are the part a plain summary silently folds into the narrative — an argument that ran fifteen minutes and ended in "let's come back to it" gets one neutral sentence and vanishes. The third section catches the numbers people say out loud with confidence and nobody checks, which is where a surprising share of bad decisions start.

1. Paste the transcript raw

   Do not tidy it first. Interruptions, half-sentences, and someone changing their mind mid-paragraph are signal — the difference between a decision and a suggestion is usually in exactly that mess.

2. Give the roles if you have them

   Ownership is the field that goes wrong most often. Without roles the model assigns an action to whoever spoke last about it, which is frequently the person least able to do it.

3. Read section two first

   Section one is the record. Section two is the work. If section two is empty in a meeting that ran an hour, either the meeting was unusually decisive or the transcript is incomplete.

4. Send it back for what it missed, not for a rewrite

   If a decision is misattributed, say so and ask it to recheck that one item against the transcript. Asking for the whole thing again tends to produce a differently-wrong version.

• When the transcript is auto-generated and speakers are unlabelledOwner attribution becomes guesswork, and a confident wrong owner is worse than none. Either label the speakers first, or run it and treat every owner as unverified.
• When the meeting was mostly a decision made before it startedSome meetings exist to announce, not to decide. The open-items section will be thin and that is accurate — do not read the thinness as the prompt failing.
• For anything that will be relied on as a formal recordBoard minutes, disciplinary meetings, and regulated processes have their own rules about what a record is and who approves it. This produces working notes, not minutes.

• BusinessRagnarökExplain a contract clause in plain languageRewrites a clause in plain language, then separately lists the actions it permits the other party to take — including the ones that are technically allowed but rarely expected.
• WritingApolloRemove AI tells from a draftRewrites text that reads as machine-written, then lists the specific phrases and structures that gave it away so you stop producing them.
• LearningApolloLearn a topic and get quizzed on itExplains something at three levels of detail, then questions you on it and reports which specific part you did not actually have.
• ResearchRagnarökCompare options and see where they differCompares the options on the things that would actually change your decision, and says separately which differences do not apply to you.