Skip to content

Getting Started with Jac#

Welcome to Jac - a programming language designed for the AI era. Jac extends Python with powerful new paradigms while maintaining full compatibility with the Python ecosystem.

What is Jac?#

Jac is a Python superset designed for the AI era, introducing three key innovations:

  1. One Language for Full-Stack: Build complete web applications - React-style frontends and Python-powered backends - in a single language with seamless integration
  2. AI-First Design: Native LLM integration via the by llm() syntax - no prompt engineering required
  3. Scale-Native Execution: Write once, run anywhere - from local development to cloud deployment without code changes

Jac also introduces new abstractions like Object-Spatial Programming (OSP) for working with graph-based data structures using nodes, edges, and walkers as first-class language constructs.

Quick Start#

1. Install Jac#

# Create a virtual environment (recommended)
python -m venv jac-env
source jac-env/bin/activate  # Linux/Mac
# jac-env\Scripts\activate   # Windows

# Install Jac
pip install jaclang jac-client jac-byllm jac-scale

# Verify installation
jac --version

2. Install the VS Code Extension#

For the best development experience with syntax highlighting, autocomplete, error detection, and graph visualizations:

VS Code: Open Extensions (Ctrl+Shift+X), search "Jac", and install the official Jac extension.

Cursor: Download the latest .vsix from GitHub releases, then use Ctrl+Shift+P → "Install from VSIX".

3. Hello World#

Create a file named hello.jac:

with entry {
    print("Hello, Jac World!");
}

Run it:

jac run hello.jac

Output:

Hello, Jac World!

4. Create a Project#

Create a full-stack project with frontend and backend:

# Create a new project with client-side support
jac create --use client myapp
cd myapp

# Start the development server (uses main.jac by default)
# If main.jac doesn't exist, specify your entry file: jac start app.jac --watch
jac start --watch

Open http://localhost:8000 to see your app running. The --watch flag enables Hot Module Replacement - edit your code and see changes instantly!

Note: If your project uses a different entry file (e.g., app.jac), you can specify it: jac start app.jac --watch.

Your project includes:

  • main.jac - Your application code (frontend + backend in one file)
  • jac.toml - Project configuration
  • Ready-to-use React-style components with JSX syntax

Choose Your Path#

Jac supports multiple development styles. Choose the path that fits your project:

🖥️ CLI & Scripts#

Build command-line tools, automation scripts, and backend services.

jac run myapp.jac

Next: Syntax & Basics → Object-Spatial Programming

🌐 Full-Stack Web Apps#

Most Popular Starting Point

Build complete web applications with React-like frontends and Jac backends - one language for everything.

# Create a full-stack project
jac create --use client myapp
cd myapp

# Start development server with hot reload
jac start --watch

Open http://localhost:8000 to see your app. Edit main.jac and watch it update instantly!

Key Features:

  • cl { } blocks for React-style components with JSX
  • has variables automatically become reactive state
  • root spawn MyWalker() calls backend directly - no fetch() boilerplate
  • Hot Module Replacement (HMR) for instant feedback

Next: Full-Stack Guide

🤖 AI Integration#

Build LLM-powered applications with native AI syntax - no prompt engineering required.

import from jac_cloud.core.llm { llm }

can summarize(text: str) -> str by llm();

with entry {
    summary = summarize("Long article text here...");
    print(summary);
}

Next: AI Integration Guide

🚀 Production Deployment#

Deploy to Kubernetes with zero configuration changes.

# Deploy to Kubernetes
jac start myapp.jac --scale

# Remove deployment
jac destroy myapp.jac

Next: Production Guide

Learn More#

Resource Description
Installation Guide Detailed setup with IDE configuration
Introduction to Jac Core concepts and philosophy
Quickstart Tutorial Build a complete application
The Jac Book Comprehensive learning resource

Who is Jac For?#

  • Startups: Rapid prototyping with one language for frontend, backend, and AI
  • AI/ML Engineers: Build LLM agents and agentic workflows naturally
  • Full-Stack Developers: Modern language features with Python ecosystem access
  • Students: Approachable on-ramp to AI development

When to Use Jac#

Jac excels when:

  • Your problem domain involves connected data (social networks, knowledge graphs)
  • You need LLMs deeply integrated into your application logic
  • You want to reduce DevOps overhead with auto-generated APIs
  • You prefer cleaner syntax with mandatory typing and modern features