GitHub - unbody-io/unbody: The Supabase of AI era. A modular, open-source backend for building AI-native software — designed for knowledge, not static data.

2 min read Original article ↗

Unbody

Unbody is building the Supabase of the AI era.

Building AI features today is a complex and fragmented process. Unbody fixes that. We're continuing our journey out in the open. The project is in early development, so expect some rough edges.

Getting Started

Note: These steps provide a basic setup. For more detailed guidance, join our Discord community.

Prerequisites

  • Node.js LTS (20 or 22)
  • Docker and Docker Compose
  • yarn (npm won’t install dependencies correctly)
  • OpenAI API key

Installation

# Clone the repository
git clone https://github.com/unbody-io/unbody

# Navigate to the project directory
cd unbody

# Install node modules
yarn

Environment Setup

# Create .env.local from template
cp .env.example .env.local

# Edit with your preferred editor, add your OpenAI API key
vim .env.local

Running the Application

# Start the required services
docker compose up -d

# Start the application
yarn start

Create Your First Project

We've prepared a demo project for you to quickly get started:

# Clone the examples repository
git clone https://github.com/unbody-io/examples.git
cd examples

Add a Data Source

# Add the storage/ directory from the examples repo as a source to Unbody
yarn unbody-cli source add

Monitor Indexing Progress

Head over to the Temporal dashboard at http://localhost:8233/ to see your files being indexed.

Note:

  • The dashboard needs to be manually refreshed to get the latest state
  • Once indexing is complete, follow the README from the examples repository

If you run into any issues during the setup process, or would like to give us any kind of feedback, join our Discord community and we'll be happy to answer any questions.

Links