Animated data flow diagram

AI Meeting Assistant: Real-time Transcription & Insights

Version: 1.0.0 | Last Updated: 2025-05-16

Integrates with:

OpenAI Recall.ai Supabase PostgreSQL

Overview

Unlock Real-time Meeting Intelligence with this AI Agent

Stop manually transcribing meetings and struggling to recall key decisions. This AI Meeting Assistant automates the entire process, acting as your dedicated note-taker and analyst for any online call. It uses Recall.ai to seamlessly join meetings (like Google Meet, Zoom) and capture live transcriptions via AssemblyAI. Then, it leverages an OpenAI Assistant to understand the conversation, identify important points based on keywords, and automatically generate notes. All transcriptions and AI-generated notes are systematically stored in your Postgres/Supabase database, linked to the specific meeting.

This agent empowers you to focus on the discussion, confident that every crucial detail is captured and organized for easy access and review.

Key Features & Abilities:

  • Automated Meeting Participation: Deploys a Recall.ai bot to join specified online meetings.
  • Real-time Transcription: Captures live speech-to-text using AssemblyAI (via Recall.ai) for high accuracy.
  • AI-driven Conversational Analysis: Utilizes an OpenAI Assistant to process the transcript, understand context, and respond to triggers.
  • Intelligent Note-Taking: If specific keywords (e.g., "Jimmy", customizable) are mentioned, the AI Assistant is prompted to take relevant notes using a custom tool that writes directly to your database.
  • Structured Data Storage: Organizes meeting metadata, full transcripts, and AI-generated notes in a PostgreSQL/Supabase database for easy querying and future use.
  • Persistent AI Context: Maintains an OpenAI Assistant thread for each meeting, allowing for ongoing, context-aware interactions if extended.

Use Cases

  • Automated generation of meeting minutes and action items for B2B SaaS project teams.
  • Real-time call analysis for sales and support reps in e-commerce, capturing customer pain points and commitments.
  • Archiving verbatim customer feedback from discovery calls for SaaS product development.
  • Ensuring compliance and record-keeping for important client meetings.
  • Building a searchable knowledge base from all internal and external team meetings.

Prerequisites

  • An n8n instance (Cloud or self-hosted).
  • Recall.ai API Key.
  • OpenAI API Key with access to the Assistants API (e.g., GPT-4, GPT-3.5-turbo).
  • Credentials for your PostgreSQL database (can be a Supabase Postgres instance).
  • Credentials for your Supabase project (if using the Supabase node specifically, though Postgres credentials might suffice if it's the same DB).
  • A PostgreSQL/Supabase database with the public.data table created. Schema:
    CREATE TABLE public.data (
      id uuid NOT NULL DEFAULT gen_random_uuid(),
      date_created timestamp WITH time zone NOT NULL DEFAULT (now() AT TIME ZONE 'utc'::text),
      input jsonb NULL,
      output jsonb NULL,
      CONSTRAINT data_pkey PRIMARY KEY (id)
    );
    

Setup Instructions

  1. Download the n8n workflow JSON file.
  2. Import the workflow into your n8n instance.
  3. Configure Credentials & Endpoints:
    • Recall.ai Bot Creation: In the 'Create Recall bot' (HttpRequest) node:
      • Update the Authorization Header under 'Authentication' with your Recall.ai API Key (Bearer Token).
      • In the node's 'Body Parameters', modify the destination_url value. Replace the placeholder URL with the production webhook URL from your 'Scenario 2 Start - Webhook' node (copy it after importing and activating the webhook node once).
    • OpenAI Thread Creation: In the 'Create OpenAI thread' (HttpRequest) node, select your OpenAI API credential under 'Authentication'.
    • OpenAI Assistant: In the 'OpenAI1' node, select your OpenAI API credential. Ensure the 'Assistant ID' parameter (asst_D5t6bNnNpenmfC7PmvywMqyR) is an assistant you have access to or replace it with your own.
    • Supabase Data Record: In the 'Create data record' (Supabase) node, select or create your Supabase API credentials.
    • PostgreSQL Nodes: In 'Insert Transcription Part' (Postgres) and 'Create Note' (PostgresTool) nodes, select or create your PostgreSQL credentials.
  4. Database Table: Ensure the public.data table is created in your PostgreSQL/Supabase database using the schema provided in 'Prerequisites'.
  5. (Optional) Customize Keyword Trigger: Modify the condition in the 'If Jimmy word' node to change the keyword(s) that trigger the OpenAI Assistant for note-taking.
  6. (Optional) Customize OpenAI Assistant: The workflow uses a pre-defined Assistant ID. You can create and use your own Assistant via the OpenAI platform and update the ID in the 'OpenAI1' node. Customize its instructions and tools as needed (the 'Create Note' node is set up as a tool for it).
  7. Activate the Workflow: Save and activate the workflow.
  8. Initiate a Meeting Bot: To start, manually run 'Scenario 1' by executing the 'Scenario 1 Start - Edit Fields' node (you can update the meeting_url here) or trigger it via its own webhook/API call if you adapt it. This will create the Recall.ai bot for the specified meeting.

Tags:

AI AgentAutomationOpenAIMeeting AutomationTranscriptionProductivityRecall.aiSupabasePostgreSQL

Want your own unique AI agent?

Talk to us - we know how to build custom AI agents for your specific needs.

Schedule a Consultation