Loading page
About this tool
Extract action items, decisions, and key points from meeting transcripts automatically.
The Meeting Transcript Action Extractor scans a pasted meeting transcript line by line and sorts it into action items, decisions and key points using cue-phrase patterns such as "action item", "todo", "follow up", "decision", "agreed", "decided" and "key point". It also pulls out @mentions as owners and phrases like "deadline:", "due by" or "by Friday" as due dates, so a raw call log becomes a structured follow-up list. Meeting notetakers, scrum leads and anyone who records calls get a copyable summary and a downloadable .txt report in one pass.
Open Meeting Transcript Action Extractor on AltFTool — it loads instantly in your browser.
Paste your transcript into Paste Meeting Transcript, one utterance per line, or press Load Sample to load the built-in sprint-planning example.
Press Extract Actions; each line is tested against action cues (action item, todo, follow up, responsible, [x]) first, then decision cues (decision, agreed, decided, concluded, going with), then key-point cues, so it lands in exactly one bucket.
Stat cards count Action Items, Decisions, Key Points and Owners, each item card carries its @owner and deadline badges, and Copy Summary or Download Report saves the result as meeting-summary.txt.
Each line is classified as an action, a decision or a key point instead of being dumped into a single summary paragraph.
@mentions become owner tags and date phrases become deadline tags on the individual action item they appeared with.
The same transcript always yields the same output because extraction is pattern matching, so nothing is paraphrased or invented.
It matches each line against action cue patterns: "action item", "todo", "follow up", "responsible", a [x] checkbox, an assignment phrase, or a modal like will / need to / must / should. A line is tested for actions first, then decisions, then key points, so each line lands in exactly one bucket.
No. The regex matching, owner detection and summary formatting all run in your browser tab, and the downloaded meeting-summary.txt is generated locally from a blob, so confidential call content never leaves the device.
Keep one utterance per line and use explicit cue words. Prefixing tasks with "Action item:", decisions with "Decision:" and owners with @handle gives the cleanest result; lines longer than 20 characters that match no cue and are not header lines fall through to key points.
Yes, if you paste the plain-text export. Speaker-labelled exports work because classification is per line; timestamps are simply carried along as part of the line text, and you can strip them first if you want tighter output.