TABLE OF CONTENTS▼

ASTACK
DEVELOPER
DOCUMENTATION

Build real-time video-to-video AI applications with sub-1000ms response times. AStack provides production-ready APIs for conversational AI with voice, text, and video input.

QUICK START

Get up and running in 5 minutes

CLIENT SDK

Browser SDK for real-time sessions

SERVER SDK

Backend SDK for session management

WEBSOCKET

Real-time protocol reference

API REFERENCE

REST API, auth, and rate limits

SUPPORT

Get help from our team

ARCHITECTURE

HOW IT WORKS

1
Your backend creates a session using the Server SDK and generates a session token
2
Your frontend connects to the worker using the Client SDK with the session token
3
AStack spawns a dedicated GPU container for real-time voice, text, and video AI

1:1 CONTAINER MODEL

Each session gets its own dedicated GPU container running ASR, LLM, TTS, and Audio2Face inference. This guarantees consistent low latency and full isolation between sessions.