Summary

W-Next AI Drawio is a Next.js application integrating AI with draw.io for natural language-driven diagram creation and modification. It supports a wide range of large language models, including popular domestic Chinese options, making it particularly useful for Chinese developers seeking to streamline diagramming workflows.

How It Works

Utilizing the Vercel AI SDK and react-drawio within Next.js, the app processes natural language prompts via a chat interface. AI models generate or modify draw.io's XML format, enabling dynamic diagram creation, modification, and replication from images or documents.

Quick Start & Requirements

• Try Online: A demo is available at https://next-ai-drawio.jiang.jp/, though self-hosting is recommended for better model choices (e.g., Claude Sonnet 4.5/Opus 4.5).
• Docker: Recommended for local use. Run with docker run -d -p 3000:3000 -e AI_PROVIDER=openai -e AI_MODEL=gpt-4o -e OPENAI_API_KEY=your_api_key ghcr.io/wangfenghuan/w-next-ai-drawio:latest. Configure AI providers via environment variables.
• Local Install: Clone repo, npm install, configure .env.local with AI provider details.
• Prerequisites: Node.js/npm, Docker (optional), and AI provider API keys.
• Model Requirements: Requires LLMs adept at structured text generation (draw.io XML). Claude, GPT-5.1, Gemini 3 Pro, DeepSeek V3.2/R1 are recommended.

Highlighted Details

• Multi-Provider Support: Integrates with OpenAI, Anthropic, Google AI, Azure OpenAI, Ollama, OpenRouter, DeepSeek, SiliconFlow, Minimax, AWS Bedrock.
• Advanced Diagramming: Natural language creation, image replication, PDF/text import, animated connectors, and specialized cloud architecture diagrams (AWS, GCP, Azure).
• AI Interaction: Interactive chat for real-time refinement and AI reasoning display.
• Version Control: Comprehensive diagram history.

Maintenance & Community

Maintained by wangfenghuan, with sponsorship opportunities via GitHub. Support via GitHub issues or email (me[at]jiang.jp).

Licensing & Compatibility

Licensed under Apache 2.0, permitting commercial use and modification under its terms.

Limitations & Caveats

Self-hosting is advised for optimal performance and model selection. Users must provide their own API keys. Setting an access code (ACCESS_CODE_LIST) is crucial to prevent unauthorized usage and token depletion. Effectiveness depends heavily on the chosen LLM.