PharmaGrow Voice Agent
From AI Studio to Working MVP
The exact steps to wrap the Gemini Live SDK code into a browser-accessible voice agent for Priya.
11 steps·~2 hrs to working MVP·Node.js + Gemini Live
The code AI Studio gave you is a Node.js script— it runs on a server, not in a browser. It connects to Gemini Live via WebSocket, sends a text message, gets an audio response, and saves it as a WAV file. That’s the core engine. Your MVP needs to take this engine and wrap a browser UI around it so a client can open a webpage, click a button, speak, and hear Priya respond.
Ready to write the code?
Say the word and the complete server.ts and index.html files — with the system prompt and knowledge base already included — will be generated and ready to run.