LIVESTACK

A full-stack, realtime AI framework for JavaScript developers.

The future of AI isn't passive chatbots — it's realtime applications that are always on, always listening, always ready to help. Livestack makes building these apps approachable, from concept to deployment in hours.

Powered by a stream & liveflow architecture, Livestack handles intensive realtime AI workloads with graph-based orchestration, built-in persistence, and auto-scaling — so you focus on the idea, not the infrastructure.

Quick install

npm install @livestack/core @livestack/gateway @livestack/client

Key Features

Liveflow Architecture

Graph-based realtime workflows with multiple input/output streams per node — built for visual inspection and debugging.

Full-Stack Scaffolding

Bootstrap and deploy new AI projects with minimal configuration. Frontend, backend, and AI pipeline — all wired up.

Any Modality

Process text, image, audio, and video through unified streaming pipelines. Mix and match modalities freely.

Auto-Scaling

Provisioning system that scales your AI workloads across machines with zero configuration. Hybrid local and cloud deployment.

Built-in Data Logging

Every in-app activity is persisted and validated out of the box — ready for inspection, replay, and model training.

Dev-Friendly

Designed for JavaScript developers. Reasonable defaults, minimal surprises, and new abstractions only when necessary.

How Livestack Compares

vs. LangChain, LlamaIndex, ComfyUI, Ray

FeatureLivestackOthers
Realtime workflows (Liveflow)
Full-stack app scaffolding
Job auto-scaling
Any modality I/O~
Multiple I/O per node~
Streaming & async I/O~
Long-running stream processing~
Cloud + on-prem hybrid
Out-of-the-box data logging

Design Principles

  • Simplicity over complexity — reduce cognitive load as much as possible.
  • Reasonable defaults — minimal configuration burden.
  • Minimal surprises — new abstractions only when absolutely necessary.
  • Battle-tested — every feature is stress-tested before release.