Inspiration

Every Twitch streamer knows the feeling: chat goes silent, energy drops, and viewers start to leave. Streamers can't watch chat, play the game, and be a data analyst at the same time. We built StreamForge to solve Twitch's biggest growth problem: keeping the conversation alive.

What it does

StreamForge is an AI co-pilot that listens to Twitch EventSub in real-time. It visualizes chat energy and acts like a producer in your ear. When it detects a sustained lull in chat (fewer than 5 messages per minute for over 30 seconds), it uses OpenAI (gpt-4o-mini) to generate a single, hyper-specific engagement prompt based on recent chat context. It is delivered privately to a dashboard.

How we built it

  • Backend: Node.js, Express, Socket.io
  • Twitch Integration: @twurple/eventsub-ws for real-time WebSocket chat listening.
  • AI Brain: OpenAI Node.js SDK (gpt-4o-mini).
  • Frontend: A custom dark-mode dashboard served directly by Express.
  • Deployment: Render

Challenges we ran into

The biggest challenge was building and deploying an entire full-stack application, including a live cloud deployment, using only an Android mobile phone. Due to mobile limitations, running a local development environment or using the desktop Codex app was impossible. I had to architect the code using the ChatGPT interface, manage GitHub entirely through the mobile web browser, and configure the Render deployment without a computer.

Accomplishments that we're proud of

We are incredibly proud that we successfully deployed a live, working web application to the internet from a mobile phone. We successfully generated the backend architecture for Twitch EventSub and OpenAI integration.

What we learned

We learned how to manage Twitch EventSub WebSockets, handle asynchronous AI API calls, and deploy Node.js applications to cloud platforms like Render using only mobile browser tools.

What's next for StreamForge

Adding full OAuth integration so streamers can log in with their Twitch accounts, adding a live OBS overlay to welcome new chatters, and expanding the AI prompt generation to understand the specific game being played.

Share this project:

Updates

Submission history