For AI Agents

Build with AI, Ship with Flox

Move fast with AI tools. Flox gives you batteries-included environments for rapid prototyping and the guardrails to ship AI-generated code to prod.

From Zero to Hero with Flox

0-30 min

Quick Start

Set up AI tools and start experimenting with new ideas in minutes.

30 min-4 hrs

Build with AI

Rapid prototyping, AI pair programming, and GPU-accelerated development.

4-8 hrs

Ship It

Deploy your creation to production and show it to the world.

5 min

Set Up MCP

You want your AI coding assistant to help you build. Run `flox activate`, set up your tool of choice (Claude Code, Cursor, Codex, etc.). Now your AI assistant can understand your environment and generate code that actually runs. No more 'install this package' when it's already installed. The AI knows what you have.

30 min

Try an Experiment

Shower thought: 'What if I built this with Bun instead of Node?' You spin up a Flox environment, install Bun and SvelteKit, scaffold a project. Try it out. It's interesting but not quite right. Delete the environment. Your main system? Still clean. Total time invested: 20 minutes. Zero regrets.

2 hours

Add Dependencies

You're building a real-time chat app. Your AI assistant suggests using Redis for real-time message pub/sub. 'Can you set that up?' you ask. The AI uses Flox to install Node, PostgreSQL, and Redis. Generates working code. You run it. It actually works. This is the future.

4 hours

Use your GPU

You want to fine-tune a model for your side project. Your Flox environment has CUDA, PyTorch, and Jupyter ready to go. Load your dataset, start training. GPU goes brrrr. The model converges. You didn't spend an hour debugging CUDA dependencies—you spent it on your actual idea.

8 hours

Demo Time

Hackathon ends in 3 hours. Your app works locally. You deploy the same Flox environment to Kubernetes—no Dockerfile needed. Ship it. It works. Judges are impressed. You win. Your teammate asks 'how did you deploy so fast?' You just smile and say 'Flox.'

How It Works

1. Set Up Your AI Coding Environment

Install your AI coding tool of choice (Claude Code, Cursor, Codex, Copilot, etc.) and integrate it with Flox. Your AI assistant can now understand and work with your environment.

# Initialize a Flox environment
flox init -d my-project
flox install -d my-project claude-code flox/flox-mcp-server

# Activate the environment
flox activate -d my-project

# Connect your AI tool
claude mcp add flox -- flox-mcp

2. Let AI Help You Build

Ask your AI assistant to set up your stack. It can install packages, configure environments, and even write code—all within your Flox environment.

claude "set up a Next.js + Postgres project"

# Your AI can execute commands like:
flox install nodejs postgresql
npm create next-app@latest my-app

3. Share With Your Team or AI Agents

Push your environment to FloxHub. Other developers and AI agents can activate the exact same setup instantly.

# Push to FloxHub
flox push

# Your team or other agents can activate the environment
flox activate -r your-username/my-project

4. Deploy to Production Safely

Deploy your Flox environment to Kubernetes without building Docker images. Full traceability and easy rollback for AI-generated code.

# Deploy to K8s with Flox runtime
apiVersion: v1
kind: Pod
metadata:
  name: my-ai-app
  annotations:
    flox.dev/environment: "your-username/my-project"
spec:
  runtimeClassName: flox
  containers:
    - name: app
      image: flox/empty:1.0.0
      command: ["npm", "start"]