Module @livestack/tutorials

<h1 align="center">LivestackΒ πŸ¦“ </h1>

<p align="center"> <img src="https://img.shields.io/github/license/zigzag-tech/livestack?style=flat-square" alt="GitHub license" /> <a href="https://www.npmjs.com/@livestack/core"><img src="https://img.shields.io/npm/v/@livestack/core.svg?style=flat-square" alt="npm version" /></a> </p>

<p align="center">A full stack, realtime AI framework for JavaScript developers. <br />From concept to deployment in hours. </p>

<p align="center"> <img src="_misc/assets/intro.gif" width="70%" height="auto" /> </p>

We believe the future of AI is not just about passive chatbots. It's about realtime AI applications that are always on, always listening, always ready to help.

At present, however, to implement such an AI project is still a complex task. Existing AI frameworks aren't well-suited for realtime interaction and often require extensive program design and coding. Once it's prototyped, there are still many steps to take to scale and deploy the application before it can be used by others.

Livestack, designed from the ground up, aims to democratize this process, making it approachable not just for seasoned developers but also for those with basic coding experience, and eventually, for everyone.

We look forward to seeing your next idea come to life with Livestack!

# Install both backend and frontend packages
npm install @livestack/core @livestack/gateway @livestack/client

For a detailed walkthrough, choose one of the tutorials from the quickstart section below.

Discord Channel

Livestack aims at providing:

  • Stream & liveflow based architecture.
  • Tools and patterns for realtime orchestration.
  • Simple, hassle-free bootstrap & deployment process for new AI project.
  • Graph-based liveflows for better reasoning, visual inspection and debugging:
    • Support for multiple input & output streams for every processing node
    • Visualization for easy inspection and debugging (as part of Livestack Cloud).
  • Built-in data persistence & validation of all in-app activities for easy inspection and model training.
  • Provisioning system that automatically scales your AI project to many machines with zero configuration.
  • Support for hybrid local and cloud deployment.
  • (WIP) full-stack realtime AI compnents, patterns and guidelines for rapid prototyping, including:
    • live speech transcription.
    • live RAG liveflow.
    • Game development
    • realtime image generation.
Features Livestack Langchain/<br />LangGraph Llamaindex ComfyUI Ray.io
Liveflow (realtime workflow) βœ… ❌ ❌ ❌ ❌
Full-stack app scaffolding βœ… ❌ ❌ ❌ ❌
Job autoscaling βœ… ❌ ❌ ❌ ❌
Any modality as input/output <br />(e.g. text/image/audio) βœ… ❌ ❌ ❌ βœ…
Allow any number of input/output <br />per processing node βœ… ❌ ❌ βœ… βœ…
Streaming & async input/output βœ… πŸ”Ά πŸ”Ά ❌ πŸ”Ά
Long running stream processing βœ… ❌ ❌ ❌ βœ…
Cloud+on-prem mixed deployment <br />(local + cloud hybrid) βœ… ❌ ❌ ❌ πŸ”Ά
Out-of-the-box data logging βœ… ❌ ❌ ❌ ❌
Python + JS Hybrid language liveflows 🚧 βœ… ❌ ❌ ❌
Official support audio Input & output 🚧 ❌ ❌ ❌ ❌
Official support for Image as output 🚧 ❌ ❌ βœ… ❌

Legend:

  • βœ…: Fully supported
  • πŸ”Ά: Partially supported
  • ❌: Not supported
  • 🚧: Work in progress
  • Simplicity triumphs over complexity: reduce user's cognitive load as much as possible.
  • Reasonable defaults to reduce burden of configuration.
  • Minimal surprises and gotchas.
  • Introduce new concepts and abstractions only when absolutely necessary.
  • Battle-test a new feature, hurt it plenty, before releasing it.