A real-time, multimodal AI assistant built with React and the Google GenAI SDK. Jarvis integrates live audio and video streaming with advanced AI capabilities, including real-time conversation, internet search, and image generation, powered by Gemini 2.5 Flash and Nano Banana Pro (Gemini 3 Pro Image Preview).

Watch Demo Video

Try the app in AI Studio:

(Note: Requires a billing account for Nano Banana Pro usage)

• Architecture: Overview of the technical stack, core services (LiveService, ToolService), and data flow.
• API Keys & Configuration: Instructions for setting up the app with Gemini Developer API (AI Studio) or Vertex AI (Google Cloud).
• Prompt: The original system prompt used to bootstrap the application.

• How to use Gemini Live API with Native Audio in Vertex AI

The app uses the Live API with Search grounding and image generation/reimagination capabilities. Here are some example prompts to try:

• "Hello Jarvis, can you tell me today's weather in San Francisco?"
• "Jarvis, can you create a photo of a futuristic city skyline at sunset?"
• "Jarvis, please take a photo of me and reimagine it as if I'm in a castle."

Prerequisites: Node.js

1. Install dependencies:

   npm install

2. Configure Environment: Set the GEMINI_API_KEY (or Vertex credentials) in .env.local. See API Keys & Configuration for details.

3. Run the app:

   npm run dev