packages/twenty-companion/README.md
WARNING: This application is a Proof of Concept (POC) and must NOT be used in production. It is intended for demonstration and experimentation purposes only. Security, stability, and performance have not been validated for production use.
This is a demo application that shows off what you can build with the Recall.ai Desktop Recording SDK.
This repo is intended to be a mockup of the kind of experience you can build using the Desktop Recording SDK.
Need help? Reach out to our support team [email protected].
Copy the env.example file to a .env file:
cp .env.example .envReplace RECALLAI_API_URL with the base URL for the Recall region that you're using that matches your API key, example:
RECALLAI_API_URL=https://us-east-1.recall.aiModify .env to include your Recall.ai API key:
RECALLAI_API_KEY=<your key>Required: This project also uses live transcription with Assembly AI. You'll need to configure your own Assembly credentials on the Recall.ai dashboard. Follow our AssemblyAI real-time transcription guide to set this up.
If you want to enable the AI summary after a recording is finished, you can specify an OpenRouter API key.
OPENROUTER_KEY=<your key>
To automatically create callRecording records in Twenty when a meeting starts (and mark them as ended when the meeting closes), configure:
TWENTY_API_URL=http://localhost:3000
TWENTY_API_KEY=<your key>
The call-recording Twenty app must be installed in your workspace first (packages/twenty-apps/internal/call-recording). Generate an API key at <your-twenty-instance>/settings/api-webhooks.
To launch the Twenty Desktop application, start the server first, then the app:
npm ci
npm start