Meeting Action Item Extractor

WHAT IT DOES

This tool reads meeting content from any webpage—meeting minutes, call transcripts, discussion notes, agendas, or selected text—and uses AI to extract a structured list of action items. It analyzes the content and returns each action item with: task description, owner (who is responsible), due date (when it should be completed), and priority (high/medium/low). You can choose from three output formats: Bullet List (easy to scan), Table (structured columns), or Structured (clear sections). The action items appear in the panel where you can review and copy them. This tool helps meeting attendees, project managers, and team leads capture and track action items efficiently, reducing the time spent manually reviewing meeting notes.

WHERE IT RUNS

The script runs on every website and web application you visit. It works with any page containing meeting content—meeting note platforms (Notion, Confluence), call transcripts, agenda documents, discussion forums, and more. It works regardless of framework—React, Vue, Angular, Svelte, or plain HTML—and extracts content from the page as rendered in the browser.

HOW TO USE

Upon installation, a floating panel appears with the tool interface. Select the meeting content on the page (or navigate to a page containing notes), choose your preferred output format (Bullet List, Table, or Structured), and click "Scan & Extract Actions." The tool extracts the content and sends it to the AI bridge. Within seconds, the AI returns a structured list of action items. The panel shows the source content preview and the extracted action items. Click "Copy Actions" to copy the action items to your clipboard, or "Copy Meeting" to copy the original text. Review the action items and add them to your task tracker.

TECHNICAL MECHANISM: INTELLIGENT MEETING CONTENT EXTRACTION WITH AI ACTION ITEM PARSING

This tool uses a multi-layer approach to extract meeting content. It first checks for user-selected text, then looks for common meeting containers (meeting, minutes, notes, transcript, agenda, discussion), then scans sections and topics. The extracted text is cleaned and formatted for the AI. The tool attempts WebMCP (navigator.modelContext) first if available—a W3C proposal for structured tool calls—falling back to the AI backend via GM_xmlhttpRequest. The AI is prompted to return a structured list of action items with task, owner, due date, and priority in the selected format. The response is parsed and displayed for review.

THE PANEL

The panel is fully draggable and remembers its position per website using GM_setValue. It automatically snaps to the nearest screen edge when released for a clean workspace. Click the collapse button to minimize it to a small floating bubble that takes minimal space. Click the bubble to expand the panel back. The panel can be resized by dragging any edge or corner. Press the Escape key to close the panel temporarily, and click the extension icon to reopen it.

PLEASE NOTE

This script requires the free BotGentz extension and an active AI subscription. Each action item extraction counts against your monthly AI usage cap. AI-generated action items should be reviewed before use—the AI may occasionally miss tasks or misassign owners. As websites frequently update their structure, scripts may need occasional updates to maintain content extraction accuracy. All data processing happens locally except for the AI request—only the extracted meeting content is sent to the AI backend. No browsing history or personal information is transmitted or stored. This tool is designed for productivity assistance and is not an autonomous agent that acts without review.