OpenMAIC

Product Introduction

1. Definition: OpenMAIC (Open Multi-Agent Interactive Classroom) is an open-source, multi-agent AI orchestration platform designed to generate and simulate interactive, immersive learning environments. It functions as a full-stack educational technology solution, combining a Next.js frontend with a server-backed runtime for persistent course generation and delivery.
2. Core Value Proposition: It solves the critical problem of infrastructure overhead and isolated AI agent experiences by providing a one-click, collaborative multi-agent classroom simulator. Its primary value is enabling developers, researchers, and educators to rapidly prototype, experiment with, and deploy complex multi-agent AI systems for education without managing disparate tools or complex setups.

Main Features

1. Agent-Powered Course Generation: The platform uses a multi-agent system to autonomously plan, build, and revise entire educational courses from a single prompt or uploaded materials (documents, audio, video). It employs a chat-first "Agent Workbench" where an orchestrator agent manages sub-agents for curriculum design, content creation, and iterative refinement, leveraging configured Large Language Models (LLMs) like GPT, Claude, or Gemini.
2. Multi-Agent Interactive Classroom Simulation: Generates a dynamic virtual classroom populated with AI teacher and AI student agents. These agents engage in real-time, voice-enabled discussions, draw on a shared digital whiteboard, and guide learners through generated content. The simulation is powered by Text-to-Speech (TTS) providers like VoxCPM2 (with voice cloning) and speech recognition via ASR providers like FunASR or OpenAI Whisper.
3. Rich, Exportable Content Types: The platform generates diverse, interactive learning modules. This includes editable PowerPoint slides (.pptx), quizzes with persistent state, interactive HTML simulations (3D visualizations, games, mind maps), and structured Project-Based Learning (PBL) activities. All content can be exported for offline use or further editing.
4. Provider-Neutral & Extensible Architecture: OpenMAIC is designed to be model-agnostic. It supports a wide array of LLM providers (OpenAI, Azure OpenAI, Anthropic, Amazon Bedrock, Google Gemini, local Ollama/Lemonade), TTS/ASR services, image generation, and search providers (Brave, SearXNG). It features a pluggable persistence stack, with a reference implementation using PostgreSQL for server-backed, durable course sessions.
5. OpenClaw Integration & SDK: Offers deep integration with OpenClaw, allowing users to generate classrooms directly from messaging apps like Feishu, Slack, and Telegram. It also provides a published npm SDK (@openmaic/* family) for developers to build upon its Domain-Specific Language (DSL), renderer, and importer tools.

Problems Solved

1. Pain Point: The significant technical barrier and time investment required to set up and orchestrate multiple AI agents for creating cohesive, interactive learning experiences. Traditional methods involve stitching together separate APIs for chat, speech, content generation, and simulation.
2. Target Audience: AI Researchers & Developers prototyping multi-agent systems; EdTech Engineers building interactive learning platforms; Educators & Instructional Designers creating dynamic course content without coding; Corporate Trainers developing immersive training simulations.
3. Use Cases: Rapid prototyping of AI tutoring systems; creating interactive product training from a PDF manual; generating debate simulations for history classes; building hands-on coding tutorials with live AI assistance; converting lecture videos into structured, interactive courses with quizzes.

Unique Advantages

1. Differentiation: Unlike singular AI tutoring tools or static content generators, OpenMAIC provides a holistic, sandboxed simulation environment. It combines course generation, multi-agent interaction, and rich media export into a single, open-source package, whereas competitors typically offer only one of these facets.
2. Key Innovation: Its "Durable Session" architecture for agentic course generation. The server-backed runtime allows long-running, stateful agent workflows to be paused, resumed, and steered by the user, with all context and materials persisted. This enables complex, multi-step course creation that survives browser refreshes or restarts.

Frequently Asked Questions (FAQ)

1. What is OpenMAIC and how does it work? OpenMAIC is an open-source AI platform that uses multi-agent orchestration to turn topics or documents into interactive virtual classrooms. You provide a prompt or file, and its AI agents automatically generate slides, quizzes, simulations, and facilitate real-time discussions between AI teachers and students.
2. Can I use OpenMAIC with local AI models without an API key? Yes, OpenMAIC supports local AI providers. You can use Lemonade for local LLM, image, and speech synthesis, or Ollama for local LLMs. For speech recognition, you can deploy FunASR locally, an open-source ASR server that works without API keys.
3. How do I deploy OpenMAIC for production use? OpenMAIC can be deployed via Docker Compose, which supports a production-ready stack with PostgreSQL for persistence. You configure your environment variables with API keys, build the Docker image, and use the server-persistence profile to run the app with a dedicated database for scalable, server-backed course storage.
4. What kinds of interactive content can OpenMAIC generate? The platform generates multiple content types: standard presentation slides, graded quizzes, interactive HTML modules (like 3D visualizations and games), and structured Project-Based Learning (PBL) activities—all within a simulated classroom UI where AI agents guide the learner.
5. Is OpenMAIC free to use and self-host? Yes, OpenMAIC is released under the permissive MIT license, making it free to use, modify, and self-host. Costs are associated with the AI provider APIs you choose to integrate (e.g., OpenAI, Anthropic) or the infrastructure for hosting the application and its PostgreSQL database.