OpenMAIC — Open Multi-Agent Interactive Classroom

Overview

OpenMAIC (Open Multi-Agent Interactive Classroom) is an open-source AI platform from the Tsinghua University MAIC Lab that transforms any topic, document, or subject into an immersive interactive classroom experience. Using multi-agent orchestration, it deploys AI teachers and AI classmates who can speak, draw on whiteboards, and engage in real-time discussions to teach any subject.

With approximately 20,000 GitHub stars and MIT licensing, OpenMAIC represents a novel application of multi-agent systems: rather than coding or automation, it focuses on education and collaborative learning. The platform supports over seven languages and integrates with messaging platforms like Feishu, Slack, and Telegram through OpenClaw, making it accessible to global users.

Available at github.com/THU-MAIC/OpenMAIC.

Features

• One-Click Lesson Generation: Convert any topic or uploaded document into a full interactive lesson instantly
• Multi-Agent Classroom: AI teachers and AI classmates lecture, discuss, and collaborate in real time
• Five Interactive UI Types: 3D visualization, physics simulations, educational games, mind maps, and online programming environments
• Rich Scene Types: Slide presentations, quizzes, interactive HTML simulations, and project-based learning modules
• AI Whiteboard: Agents sketch diagrams, write formulas, and solve equations step by step on a shared whiteboard
• Voice Interaction: Text-to-speech with voice cloning and speech recognition for natural voice-based learning
• Messaging Platform Integration: Generate classrooms directly from Feishu, Slack, Telegram via OpenClaw
• Multi-Format Export: Editable PowerPoint (.pptx), interactive HTML pages, classroom ZIP files, and MP4 videos
• Multi-Language Support: Chinese Simplified/Traditional, English, Japanese, Russian, Arabic, Brazilian Portuguese
• Multi-Provider LLM Support: OpenAI, Anthropic, Google Gemini, DeepSeek, Qwen, Kimi, MiniMax, Grok, and OpenAI-compatible APIs

Installation

# Prerequisites: Node.js ≥ 20, pnpm ≥ 10
git clone https://github.com/THU-MAIC/OpenMAIC.git
cd OpenMAIC
pnpm install

# Configure API keys
cp .env.local.example .env.local
# Edit .env.local with your API keys

# Start development server
pnpm dev

Docker deployment:

docker compose up -d

Vercel one-click deployment:

# Deploy via Vercel platform with Git integration

Quick Start

// Generate an interactive classroom from a topic
import { createClassroom } from "openmaic";

const classroom = await createClassroom({
  topic: "Introduction to Machine Learning",
  agents: ["teacher", "student-a", "student-b"],
  scenes: ["slides", "quiz", "whiteboard"],
  exportFormat: "html"
});

console.log(classroom.url);

Core Concepts

Multi-Agent Teaching

OpenMAIC deploys multiple agents in a classroom: a teacher agent who lectures and guides, and student agents who ask questions, discuss, and simulate peer interactions. The agents collaborate through orchestrated turn-taking and shared context.

Interactive Scene System

Lessons are composed of scenes — slides, quizzes, simulations, and whiteboard sessions. Each scene type supports different learning modalities: visual, interactive, and collaborative.

Topic-to-Classroom Pipeline

The pipeline takes raw input (a topic string, document, or syllabus), uses LLM reasoning to structure the content, generates scenes and agent behaviors, and produces a deployable interactive classroom.

Examples

Topic-Based Classroom

const classroom = await createClassroom({
  topic: "Climate Change and Renewable Energy",
  languages: ["en"],
  interactiveTypes: ["3d-viz", "game", "simulation"]
});

Document-Based Classroom

const classroom = await createClassroom({
  document: "uploaded-textbook.pdf",
  agents: ["teacher"],
  exportFormat: "pptx"
});

Pros

• ✅ Innovative application of multi-agent systems to education
• ✅ Highly interactive with 5 UI types and rich scene system
• ✅ Multi-language and multi-LLM provider support
• ✅ Easy deployment via Docker and Vercel
• ✅ Strong research backing from Tsinghua University
• ✅ MIT licensed and open-source

Cons

• ❌ Newer project with a narrower use case than general agent frameworks
• ❌ Requires API keys and stable LLM access
• ❌ Interactive features may be resource-intensive
• ❌ Smaller community and ecosystem compared to mainstream frameworks
• ❌ Voice and whiteboard features may have latency

When to Use

• AI-Powered Education: Building interactive courses or training materials
• Document-to-Classroom Conversion: Turning textbooks, research papers, or presentations into engaging lessons
• Corporate Training: Creating self-paced, AI-guided training programs
• Multi-Agent Collaboration Research: Studying how AI agents teach and learn from each other
• Language Learning: Creating immersive, interactive language lessons

Resources

• GitHub Repository
• OpenMAIC Documentation
• Tsinghua MAIC Lab