This project provides an AI-powered tool for generating presentation slides from a given theme. It targets users who need to quickly create structured and content-rich presentations, offering automated outline and per-page content generation with template support to streamline the workflow.

How It Works

The system employs a frontend-backend separated architecture, with the frontend built on Vue.js and the backend utilizing Python (FastAPI). Core AI functionality is driven by Large Language Models (LLMs) for web searching and content generation, enhanced by a GRPO (Proximal Policy Optimization) reinforcement learning approach to optimize output quality and user alignment. Content is generated page-by-page using streaming technology for real-time display, and the system can leverage user-uploaded knowledge bases or pre-defined templates for content and style.

Quick Start & Requirements

For development, start the frontend with cd frontend && npm install && npm run dev and the mock backend with cd backend/mock_api && python mock_main.py. For production, configure environment variables in .env after copying env_template.txt and run python start.py for a one-click deployment. Alternatively, Docker can be used via docker compose up. Prerequisites include Node.js for the frontend and Python for the backend, along with potential API keys for LLM services. Detailed backend setup instructions are available in backend/启动说明.md.

Highlighted Details

• Intelligent, structured outline generation from a user-provided theme.
• Real-time, streaming generation of per-page PPT content.
• Support for generating PPTs based on content from a user's knowledge base (parsing various file formats).
• A template system allowing users to choose from various designs and apply content separately.
• Flexibility to integrate different commercial or local AI models.
• Optimization of AI generation quality through the GRPO reinforcement learning method.

Maintenance & Community

No specific details regarding maintainers, community channels (like Discord or Slack), or project roadmaps are provided in the README.

Licensing & Compatibility

The backend components are described as "免版权" (no copyright/permissive), while the frontend is licensed under AGPL-3.0. The AGPL-3.0 license imposes strong copyleft requirements, meaning any modifications or derivative works of the frontend must also be made available under the AGPL-3.0 license, which may impact integration into closed-source commercial products.

Limitations & Caveats

The project lists chart support as implemented. However, support for uploading custom PPT templates and automatic annotation is still pending. The AI generation process relies on external web searches and LLM APIs, introducing potential dependencies and costs.