Every Meeting Summary Written, Every Action Item Captured. Automatically.

Part 1: Why Automate Meeting Summaries

The average knowledge worker attends 21.5 hours of meetings per week. They spend an additional 4.8 hours preparing for and following up on those meetings. A Microsoft study of 31,000 workers found that meeting time has increased 252% since 2020, and that 71% of workers say meetings prevent them from getting their work done.

Within this meeting overload, the meeting summary is one of the most universally neglected administrative tasks in business. Every meeting should produce a written record: what was discussed, what was decided, and who is doing what by when. In practice, meeting summaries are produced inconsistently, incompletely, and late, if at all.

The consequences compound. Action items assigned in a meeting but not captured in writing are forgotten at a rate that research consistently estimates at 30-40%. Decisions made in a meeting but not documented are re-litigated in the next meeting, consuming time that should be spent executing. New team members who were not in the meeting have no record of what was decided or why.

AI meeting summary automation changes this from a task that requires someone's time and attention to a process that happens automatically.

The documented productivity impact:

Otter.ai's research on organizations using AI meeting summaries found that teams recovered an average of 3.6 hours per employee per week, time previously spent on manual note-taking, writing up summaries after meetings, and chasing action items that were never clearly assigned. At a loaded labor cost of $75,000/year per employee, 3.6 hours per week represents $6,750 per year per employee in recovered productive time.

For a 50-person organization: $337,500 per year in recovered productivity from a single automation.

The action item capture rate improvement is equally significant. Organizations that implemented AI meeting summaries with automatic action item extraction and assignment tracked to their project management systems reported a 67% reduction in missed action items in the first 90 days.

What AI meeting summaries actually include:

A well-structured AI meeting summary contains six elements:

Attendees and duration, who was present and how long the meeting ran.

Meeting purpose, one sentence describing what the meeting was convened to accomplish.

Key discussion points, the main topics covered, in the order they were discussed, with the key points and perspectives raised on each. Not a verbatim transcript. A synthesis.

Decisions made, a bulleted list of specific decisions reached, written in the past tense ("The team decided to..."). Only decisions, not discussion.

Action items, a bulleted list of specific tasks, each with an owner name and a due date. Every action item has both. "John will update the pricing deck by Friday" not "someone should look at the pricing deck."

Open questions, items raised that require further information or a separate decision, with the person responsible for resolving each.

This structure takes a 60-minute meeting and reduces it to a 2-minute read that contains everything actionable.

Part 2: How to Build the Meeting Summary Automation

This pipeline connects to your video conferencing platform, transcribes every meeting automatically, generates a structured summary using Claude, distributes the summary to all attendees, and posts action items to your project management system.

The pipeline:

Meeting ends on Zoom, Teams, or Google Meet
→ Recording and transcript available (5-10 minutes after meeting end)
→ Webhook or API pull triggers n8n workflow
→ Transcript sent to Claude with meeting metadata (attendees, title, duration)
→ Claude generates: structured summary with all six elements
→ Summary emailed to all attendees within 15 minutes of meeting end
→ Action items extracted and posted to Asana, Notion, or Jira
→ Action item owners receive individual Slack notifications
→ Summary archived in Google Drive with searchable metadata

Platform setup:

Zoom: Enable cloud recording in your Zoom account settings. Zoom generates a transcript automatically for all recorded meetings. Use the Zoom webhook to trigger the n8n workflow when a recording is complete. The webhook payload includes the recording URL and transcript URL.

Microsoft Teams: Enable transcription in Teams meeting settings. Use the Microsoft Graph API to retrieve meeting transcripts. The Graph API provides meeting metadata (attendees, start/end time, title) alongside the transcript.

Google Meet: Enable transcription in Google Workspace admin settings. Transcripts are saved to Google Drive automatically. Use the Google Drive API to detect new transcript files and trigger the workflow.

The Claude summarization prompt:

The prompt provides Claude with the full transcript and meeting metadata, and instructs it to generate the six-element summary. Key prompt design decisions:

Speaker identification: the prompt instructs Claude to attribute discussion points and action items to specific speakers by name, using the speaker labels from the transcript. "Sarah mentioned that the timeline was aggressive" not "it was mentioned that the timeline was aggressive."

Action item extraction rules: the prompt includes explicit rules for what constitutes an action item. Any statement that includes a commitment ("I will," "we need to," "someone should," "let's make sure"), a person's name, and a task qualifies. Vague commitments without owners are listed in Open Questions, not Action Items.

Confidentiality handling: for meetings that include sensitive topics (HR matters, compensation discussions, legal matters), the prompt includes instructions to summarize at a higher level of abstraction. A flag in the meeting title ("CONFIDENTIAL: [topic]") can trigger a different prompt variant.

Length calibration: the prompt instructs Claude to calibrate summary length to meeting duration. A 15-minute standup generates a half-page summary. A 90-minute strategy meeting generates two to three pages.

Action item integration:

The automation extracts action items from the summary and creates tasks in your project management system:

Asana: create a task in the designated project with the action item as the task name, the owner as the assignee, and the due date as the due date. Add the meeting summary as a comment on the task.

Notion: create a new page in a designated database with task name, owner, due date, and a link to the full meeting summary.

Jira: create a new issue in the designated project with the action item as the summary, the owner as the assignee, and the due date in the due date field.

After tasks are created, the automation sends individual Slack DMs to each action item owner: "Hi [Name], you were assigned an action item from the [Meeting Name] meeting on [Date]: [Action Item]. Due: [Date]. Full summary: [link]."

The meeting archive:

Every summary is saved to Google Drive in a structured folder system: Year/Month/[Meeting Name]/. The file is named with the date and meeting title. The Google Drive folder is shared with all meeting participants.

Metadata tags enable search: meeting attendees, date, project or team, and key topics. A finance leader can search for all meetings involving a specific topic across the past year and retrieve the summaries instantly.

The weekly digest:

Every Monday morning, each team member receives a digest of all action items assigned to them that are due in the next 7 days, pulled from all meeting summaries across all their meetings. One email, all their commitments, consolidated from all their meetings. No hunting through notes or email threads.

Part 3: Get the Automation

The WorkplaceAI Meeting Summary Automation transcribes every meeting, generates a structured summary with extracted action items, distributes to attendees within 15 minutes, and posts tasks directly to your project management system.

What's included:

• Complete n8n workflow JSON, full pipeline from transcript retrieval through Claude summarization, email distribution, action item extraction, project management integration, and Slack notifications
• Platform integration guides, Zoom webhook setup, Microsoft Teams Graph API, and Google Meet Drive API configuration
• Six-element summary prompt library, base prompt plus variants for standup meetings, strategy meetings, customer calls, HR meetings, and board meetings
• Action item extraction rules, the prompt language that consistently captures commitments with owners and dates
• Project management integration guides, Asana, Notion, Jira, and Linear task creation via API
• Weekly action item digest template, the Monday morning email format that consolidates all outstanding commitments
• Google Drive archive structure, the folder hierarchy and naming convention for searchable meeting archives
• 45-minute setup guide, from download to first automated meeting summary delivered

Template workflow requires configuration. Technical familiarity with n8n assumed. Setup guide included. Questions: support@workplaceai.ai

Read the full implementation guide: workplaceai.ai