How OpenMAIC Turns Any Topic into a Fully Interactive AI Classroom

OpenMAIC is an open‑source AI‑driven platform that generates a full interactive classroom from a simple input such as a topic description or a PDF document. The system operates through a two‑stage pipeline: first, a large‑language model analyses the input and creates a structured outline; second, each outline item is transformed into rich teaching components such as slide decks, quizzes, interactive simulations, or project‑based activities.

One‑Click Classroom Generation

Users interact with a minimalist interface containing a single input box. After entering a learning goal (e.g., "Learn Python from scratch"), the platform produces a complete lesson within minutes, requiring no manual configuration. The generated content includes AI‑driven voice narration, whiteboard drawing, and highlighted key points.

Multi‑Agent Teaching Engine

The core innovation is a multi‑agent collaboration engine powered by LangGraph, which orchestrates distinct AI roles: an AI teacher delivers lectures via slides, whiteboard, and voice; AI classmates initiate discussions and pose questions; an AI teaching assistant summarizes knowledge, answers queries, and provides personalized feedback. The agents can access real‑time web search to keep information up‑to‑date and support multiple large‑model back‑ends such as Qwen, GLM, and OpenAI.

Classroom Components

Slide component : AI teacher uses spotlight and laser‑pen effects while narrating.

Quiz component : Supports single‑choice, multiple‑choice, and short‑answer formats with instant grading.

Interactive simulation component : Built with HTML, it can embed physics simulators or flowcharts for hands‑on experimentation.

Project‑based learning component : Assigns roles, milestones, and deliverables for deeper practice.

Validation at Tsinghua University

During the 2024 spring semester, the predecessor MAIC was piloted in two courses, attracting 700 students and generating over 100,000 interaction records (≈140 interactions per student). In the autumn semester the offering expanded to four courses with more than 2,000 participants, confirming high engagement and positive feedback on AI‑driven discussions and whiteboard explanations.

Cost Efficiency

The system’s operating cost is reported to be only one‑thousandth of traditional MOOC courses because it eliminates the need for live instructors and pre‑recorded video production; AI dynamically creates content based on learner progress and queries.

Open‑Source Release and Ecosystem Integration

In March 2026 the code was open‑sourced on GitHub (https://github.com/THU-MAIC/OpenMAIC). The open version retains the multi‑agent engine and adds new features to lower deployment barriers. Integration with OpenClaw enables classroom generation within over 20 chat applications (e.g., Feishu, Slack, Discord, Telegram, WhatsApp) without command‑line interaction.

Two deployment modes are offered: a hosted mode requiring only an access token, and a self‑hosted mode granting full control for developers and enterprises. Export options include editable .pptx slides and interactive .html pages for LMS integration.

Future Outlook

OpenMAIC demonstrates a new AI‑education paradigm that combines multi‑agent interaction, real‑time content generation, and low‑cost scalability, potentially transforming online education production and expanding access in resource‑limited regions.