Operations & Systems
Realtime Meeting Assistant w Postgres through Supabase
This n8n workflow uses Recall.ai to capture real-time meeting transcriptions via a Webhook. It integrates with OpenAI (Assistant) to process dialogue and generate notes. All meeting data, including transcriptions and AI-generated notes, are stored in Supabase (Postgres),…
Built · ~9 hours saved per week
Overview Brief
This workflow automates the creation of a Real-time AI Meeting Assistant. It smoothly integrates several powerful technologies to have a bot join a virtual meeting (like Google Meet or Zoom), transcribe the conversation live, and use an AI agent to provide real-time insights and perform actions based on the dialogue.
The primary value of this automation is transforming passive meeting attendance into an active, intelligent, and archived process. It eliminates the need for manual note-taking, ensures key decisions and action items are captured accurately, and creates a structured, searchable database of every conversation. This significantly enhances meeting productivity, clarity, and follow-up efficiency.
Technologies Used:
- n8n: The central orchestration platform that connects all services and executes the logic flow. - Recall.ai: A service that provides the meeting bot ("agent") to join calls and deliver the real-time transcription feed. - Supabase (PostgreSQL): A cloud database used as the persistent memory for the system. It stores meeting metadata, the full conversation transcript, and AI-generated notes in a structured JSON format. - OpenAI: Provides the core intelligence through its Assistants API.
AI Integration
The AI is implemented as an OpenAI Assistant, which acts as an intelligent agent listening to the meeting. Its integration is the workflow's most powerful feature:
1. Contextual Triggering: The AI isn't just a post-meeting summary tool. It is triggered _in real-time_ when a specific keyword (e.g., "Jimmy") is mentioned in the conversation. 2. Informed Interaction: When triggered, the AI is fed the most recent part of the meeting's transcript, allowing it to understand the current context of the discussion. 3. Agentic Actions (Tools): The AI is empowered with a custom tool: Create Note. This allows the AI to autonomously decide when and what to document as a significant point. Instead of just outputting text, it can execute a structured action—writing a note directly into the meeting's database record. This elevates the AI from a simple chatbot to a functional agent assistant.
Tags: Perplexity, Data Transformation, OpenAI, Notion, AI Assistant
Integrations: OpenAI, Postgres, HTTP, Supabase, Webhook