AI Meeting Assistant Agent: Summarize Transcripts & Schedule Follow-ups

Unlock Peak Meeting Efficiency with this AI Agent

This AI-driven n8n workflow transforms your Google Meet post-meeting process. It starts by fetching a specific Google Calendar event, then retrieves the associated Google Meet conference records and transcript from Google Drive. The transcript (converted from PDF to text) is fed into an OpenAI-powered Langchain Agent. This agent intelligently summarizes the discussion, extracts key highlights, action items, and identifies if a follow-up meeting is needed. If so, the agent uses its built-in "Create Calendar Event" tool to automatically schedule the new meeting in your Google Calendar and invite specified attendees.

Key Features & Benefits:

• Automated Transcript Retrieval: Fetches Google Meet transcripts linked to your Google Calendar events via Google Drive.
• AI-Powered Summarization & Analysis: Leverages OpenAI's function-calling capabilities (via Langchain's openAiFunctionsAgent) to provide concise summaries, attendee-specific highlights, and next steps from meeting transcripts.
• Intelligent Action Execution: The AI Agent can autonomously decide to schedule follow-up meetings based on the content of the transcript and instructions.
• Automated Calendar Scheduling: Creates new Google Calendar events with titles, descriptions, timings, and adds attendees as identified by the AI or predefined in the tool call.
• Structured Output: Provides parsed information (summary, highlights, next steps, details of created meetings) for easy use or integration into other systems.
• Customizable In-Workflow Tooling: Demonstrates how to build and utilize internal tools (like scheduling a meeting) for an AI Agent, all within the same n8n workflow.

• An n8n instance (Cloud or self-hosted).
• OpenAI API Key with access to a suitable model that supports function calling (e.g., gpt-3.5-turbo, gpt-4).
• Google OAuth2 API credentials configured in n8n for:
  • Google Calendar (Read & Write access for fetching events and creating new ones).
  • Google Meet API (Scopes: https://www.googleapis.com/auth/meetings.space.readonly for conference records, https://www.googleapis.com/auth/drive.readonly for transcript access if stored as Drive files by Meet). These are used via HTTP Request nodes.
  • Google Drive (Read access, to download transcript files).
• A target Google Calendar Event ID for a past meeting that was recorded on Google Meet and had transcription enabled. The transcript must be available in the Google Drive of the authenticated user.

1. Download the n8n workflow JSON file.
2. Import the workflow into your n8n instance.
3. Configure Credentials: a. 'OpenAI Chat Model1' node: Select or create your OpenAI API credentials. b. 'Get Meeting ConferenceRecords' & 'Get Meeting Transcript Location' (HTTP Request nodes): Select or create Google OAuth2 API credentials. Ensure these credentials have the necessary scopes for Google Meet API (https://www.googleapis.com/auth/meetings.space.readonly) and potentially Google Drive (https://www.googleapis.com/auth/drive.readonly) if transcripts are fetched this way. c. 'Get Transcript File' (Google Drive node): Select or create your Google Drive OAuth2 credentials with read access. d. 'Get Calendar Event', 'Create Calendar Event1', 'Add Attendee to Invite' (Google Calendar nodes): Select or create your Google Calendar OAuth2 credentials with read/write access.
4. Configure Initial Event Source: a. In the 'Get Calendar Event' node: Update the 'Event ID' parameter with the ID of a Google Calendar event that has an associated Google Meet recording and transcript (e.g., abc123). Also, select your relevant calendar in the 'Calendar' parameter.
5. Configure Calendar for New Events: a. In the 'Create Calendar Event1' and 'Add Attendee to Invite' nodes: Ensure the 'Calendar' parameter is set to the calendar where new meetings should be created (e.g., n8n-events).
6. Review AI Agent Prompt: a. Examine the 'AI Agent' node's 'Text' parameter (this is the system prompt). It dynamically pulls details from the 'Get Calendar Event' node. Adjust the instructions for the AI if you have specific summarization or action-taking criteria.
7. Tool Configuration (Workflow Self-Call): a. The 'Schedule Meeting' (ToolWorkflow node with name: create_calendar_event) is configured to call this same workflow. The 'Workflow ID' parameter is dynamically set to ={{ $workflow.id }}. This setup enables the AI agent to trigger the meeting creation part of this workflow. No changes are usually needed here unless you restructure the tool's input or the workflow's trigger mechanism for tools.
8. Activate the workflow. You can test it using the 'When clicking "Test workflow"' manual trigger after configuration.