Decisions
- • We decided to keep the three-tier layout
- • Approved the new screenshot style
Productivity
Quick answer
Paste raw meeting notes and the generator separates them into decisions made, action items with an owner and a due date, open questions and context. Anything without an owner is flagged, since unassigned actions are the most reliable predictor that a meeting produces nothing.
Paste the messy notes. Get decisions, owners, due dates and open questions sorted in one pass.
Published · Last updated
Decisions
Action items
Risks and blockers
Open questions
Markdown to paste anywhere
# Q3 pricing page kickoff ## Decisions - We decided to keep the three-tier layout - Approved the new screenshot style ## Action items - [ ] Maria will draft the new copy by Friday — **Maria**, due Friday - [ ] Sam to send the analytics export tomorrow — **Sam**, due tomorrow ## Risks and blockers - Risk: legal review is blocking the comparison table ## Open questions - Open question: do we show annual pricing by default? ## Notes - Kickoff for the Q3 pricing page
A transcript records everything and settles nothing. What a team actually needs after a meeting is four short lists: what was decided, who owes what, what is blocked and what is still an open question.
This page reads your notes line by line and sorts them using the words people naturally use — "we decided", "will send", "blocked on", "TBD". Write one thought per line and it sorts cleanly; write paragraphs and it will still try, but the buckets get noisier.
Every action item needs a name and a date. An action with neither is a wish, and it will come back to the same meeting next week untouched.
| What it answers | Notes into decisions, owners and due dates. |
|---|---|
| How the answer is produced | A meeting summary is only useful if it answers three questions: what was decided, who owes what by when, and what is still open. |
| What you need to enter | Paste your raw notes exactly as you typed them, including fragments and shorthand. |
| Where it stops being reliable | It cannot infer what was implied but never said. |
| Cost and sign-up | Free, runs in your browser, no account and no stored inputs. |
A meeting summary is only useful if it answers three questions: what was decided, who owes what by when, and what is still open. The generator restructures your notes around exactly those buckets rather than compressing everything into a paragraph nobody reads.
Decision lines are detected from commitment language — agreed, approved, will, going with — and separated from discussion, which is the noise that makes most summaries twice as long as they need to be. Action items are pulled out with their owner and any date mentioned nearby.
Anything that reads like a question, a blocker or an unresolved trade-off is routed to an open-items list. That list is usually the most valuable part of the output, because it is the section that quietly disappears from hand-written notes.
Decisions and actions get written down because someone is watching to see if they happen. Open items — the unresolved question, the disagreement nobody settled, the risk somebody flagged and then the meeting moved on — have no such enforcement, so they are the first thing to vanish from a rushed set of notes even though they are frequently the most consequential part of the conversation.
The generator treats hedging language, questions and phrases like 'we need to figure out' as a distinct category precisely because that language rarely gets written up on its own initiative. Surfacing it mechanically means the unresolved thread from minute twelve of a forty-minute call does not simply disappear because nobody thought to note it by hand.
In practice, the open-items list is what prevents the same unresolved question from being re-raised, half-remembered, three meetings later. Circulating it alongside decisions closes the loop faster than most teams manage on their own, because it puts a visible unanswered question in front of the person who can actually answer it.
A summary that lists everything said is a transcript with extra steps, and it will not be read. What people return to a summary for is a small, specific set: what was decided, what changed since last time, who owns what by when, and what is still unresolved. Anything that does not fall into one of those buckets is discussion, and discussion belongs in the recording. Applying that filter typically compresses an hour-long meeting into fifteen lines, and it is the compression, not the completeness, that makes a summary get read by the people who missed it.
Recurring meetings expose a second failure mode: the same decision gets re-litigated because nobody circulated last week's summary, so the room re-derives a conclusion it already reached. Pasting last week's notes back into the generator alongside this week's, and asking it to flag anything that contradicts a prior decision, catches the moments where a team quietly reverses itself without noticing — a scope that was cut in one meeting and casually reintroduced in the next because the summary from the first never reached the person proposing it again.
Notes mention 'need to decide on pricing tier before launch — Sarah to check with finance' buried mid-paragraph. The generator lifts this into an open item with Sarah as the implied owner, rather than letting it sit unnoticed inside three paragraphs of general discussion about the launch timeline.
A forty-minute transcript with labelled speakers produces four decisions, six actions with named owners and dates, and two open items flagged from phrases like 'we'll need to confirm scope.' Pasting the same transcript without speaker labels still extracts the decisions but leaves the actions unowned, which is the clearest sign labelling was missing.
The most valuable entry in most summaries is the open question with a named owner and a date: 'Unresolved — whether the March launch includes the mobile flow. Priya to confirm with engineering by Friday.' Without it, that question reappears in the next meeting as if it were new, and the same twenty minutes get spent again. Written down with an owner, it either gets answered or its absence becomes visible, which is the only two outcomes worth having.
Yes, though results improve substantially when speaker labels are present so commitments can be attributed.
No. Processing happens in your browser and the text never leaves your device.
Decisions and actions at the top, discussion at the bottom or omitted entirely, and a hard limit of one screen.
Both mentions are kept and flagged rather than merged, since silently picking one date could quietly commit someone to a deadline they never agreed to. Resolve the conflict in the follow-up message, not by trusting the tool to guess which speaker was right.
Written and reviewed by Jim Vernon, Editor, AI Intelligence International. Published by AI Answer Engine, a service of AI Intelligence International, and checked against our editorial standards.