🔥 IgnitionAI

IgnitionAI is an open-source organization on a mission to bring intelligent agents and reinforcement learning to the modern web.
We build modular, TypeScript-based tools and frameworks for creating AI-driven simulations, visualizations, and real-time interactions — directly in the browser or at the edge.

✨ Vision

To democratize and simplify the development of AI-powered agents and environments using:

💡 Web-native technologies (React Three Fiber, WebGL, WASM)
🧠 In-browser inference (TensorFlow.js, ONNX Runtime Web)
🧱 Modular, developer-friendly APIs
⚡ Integration with modern AI tools (LangChain.js, Vercel AI SDK)

🧩 Key Projects (coming soon)

Project	Description
`ignition-core`	Core abstractions for agents, environments, and RL loops
`ignition-r3f`	React Three Fiber integration for agent environments
`ignition-tfjs`	TF.js inference backend
`ignition-onnx`	ONNX Runtime Web inference backend
`ignition-vercel`	Vercel AI SDK integration
`ignition-langchain`	LangChain.js adapter for planning, tools, and memory
`ignition-demo-cartpole`	Starter demo: CartPole in 3D, browser-ready

🚀 Target Use Cases

Browser-based reinforcement learning environments
Educational demos & RL visualizations
Web-native AI agents with memory and planning
RAG-enabled simulations with LangChain.js
Web3, gaming, or simulation projects powered by intelligent agents

👨‍💻 Tech Stack

TypeScript / React / R3F
TensorFlow.js, ONNX Runtime Web
LangChain.js, Vercel AI SDK
Zustand, Three.js, Vite

🌍 Join the mission

We’re building the next generation of AI-enabled interfaces for the web.
Want to contribute, test, or use IgnitionAI in your project? Let’s talk.

→ https://ignitionai.fr