Recall Demo Meeting Action Items

This demo application allows users to show user-wise action items from the ongoing meeting by simply providing a meeting url.

Description:

The purpose of this demo is to show how one can easily fetch live trancripts from any meeting, and make use of the transcripts to do more advanced things. In this case, we are simply providing action items for each user in the meeting.

This demo app does 3 things:

• Uses Recall.ai API to send a meeting bot instaneously to any live meeting.
• Use the Recall.ai API to fetch meeting transcript from any live meeting.
• Use an LLM (OpenAI in this case, but you can use anything really) to provide the action items, given the transcripts.

Technologies Used

• Frontend (client-app): React, TypeScript, Tailwind CSS
• Backend (server-app): Express, TypeScript
• API: Axios for HTTP requests
• Recall.ai APIs for meeting bot and fetching transcripts
• OpenAI GPT 3.5 turbo for summarization

Project Structure

The project is divided into two main parts:

1. client-app: React application with TypeScript
2. server-app: Express server with TypeScript

Setup and Running

Pre-requisites

1. NodeJS >= v18
2. Recall.ai API Key
3. OpenAI API Key

Frontend

1. Navigate to the client-app directory
2. Install dependencies: npm install
3. Run the build command: npm run build
4. Start the client server: npm run start

The frontend will be available at http://localhost:3000 (i.e uses 3000 Port by default. However, if that Port is not available, it uses the next available Port. You can use different port by modifying PORT value in .env file.

Backend

1. Navigate to the server-app directory
2. Install dependencies: npm install
3. Run the build command: npm run build
4. Start the backend server: npm run start

The backend will be available at http://localhost:3002 . You can use another port by defining an environment variable named PORT in the server-app/.env file.

How to Use

1. cp .env.example .env
2. Populate the values in your new .env file. OPENAI_API_KEY, and RECALL_API_KEY are mandatory.
3. Build and start your client-app as mentioned above.
4. Build and start your server-app as mentioned above.
5. Open http://localhost:3000
6. Paste your live meeting link in the input box, and click on Start Recording. This adds the bot to the meeting.
7. Now, you can go and participate in your meeting.
8. The app polls for the meeting status every few seconds. Instead of polling, you might want to implement a webhook, but that is out of scope of this demo.
9. Once the meeting ends, come back to the app. It will list the user-wise action items.

PS: Please make sure you are aware of the charges from OpenAI and Recall when calling their APIs. Request to OpenAI API is only made once, when the meeting has ended.