Deepgram's Browser Agent SDK puts voice agents in web apps in four layers
- Source
- x.com
- Date
Getting a voice agent on the API is the easy part. Getting one on a web page is where teams lose a sprint. We just fixed that. 🧵
The Browser Agent SDK is live: four composable npm packages that drop a Deepgram voice agent into any web app. Widget → React components → React hooks → framework-agnostic core. Install the layer you want, the rest comes with it.
Reconnection, audio buffering, playback-aware mode tracking, KeepAlive, Silero VAD, typed events: same across every layer. The hard parts of browser voice are handled once and inherited everywhere.
Full details, six layouts, and the path from widget to vanilla JS in the blog post: https://t.co/TphMtNngUe
Context
Deepgram's Browser SDK lets a developer choose between a ready-made voice widget and progressively more control over the interface. Its architecture places React components and hooks above a shared browser core for microphone capture, connection management and audio playback. A team can customize the visible interface while retaining that underlying audio machinery.
One launch claim was subsequently corrected. Deepgram's September 18 notice says browser-side speech activity detection never shipped. Interruptions instead rely on a server event that tells the browser when the user starts speaking; the React integration uses that event to stop playback automatically.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Checking sign-in…
Loading comments…