This project provides an enterprise-grade, multi-tenant platform for developing AI applications, including agents, workflows, and Q&A systems, with comprehensive model management capabilities. It targets businesses seeking to integrate LLMs into their operations securely and efficiently, offering a modular architecture for customization and a user-friendly, low-code interface for building complex AI-driven processes.

How It Works

The platform is built on a modular architecture, enabling flexible integration of various large language models (LLMs) and business logic. Key features include a Model Hub for managing hundreds of proprietary and open-source models (supporting OpenAI API standards and multiple inference backends like vLLM, TGI), a standardized MCP (Model-Centric Programming) interface for connecting AI models to external tools, and a RAG pipeline for enterprise knowledge base construction with advanced features like multi-modal retrieval and adaptive chunking. It also offers a visual workflow studio with a low-code drag-and-drop interface and an Agent Framework supporting function calling for tool extensibility.

Quick Start & Requirements

• Installation: Docker installation is recommended.
• Prerequisites: Docker, docker compose. For source builds, development tools are needed. Specific environment variables (WANWU_ARCH, WANWU_EXTERNAL_IP) must be configured in .env.
• Resources: Requires Docker and network configuration. Specific instructions are provided for amd64 and arm64 architectures.
• Login: Access via http://localhost:8081 with default credentials admin/Wanwu123456.
• Documentation: Detailed operational manual available.

Highlighted Details

• Apache 2.0 License: Friendly for commercial use and secondary development.
• Multi-tenancy: Supports account systems for cost control, data isolation, and business scalability.
• 信创 (Xinchuang) Adaptation: Compatible with domestic Xinchuang databases like TiDB and OceanBase.
• Comprehensive Tooling: Covers model management, RAG, MCP, agents, and visual workflows.

Maintenance & Community

The project is actively developed with recent updates in August 2025. It welcomes community contributions and offers professional team support for ecosystem partners. Links to QQ groups are provided for community interaction.

Licensing & Compatibility

Released under the Apache License 2.0, which is permissive for commercial use and derivative works.

Limitations & Caveats

The README mentions specific troubleshooting steps for Linux (memory limits) and Windows (bash interpreter issues), indicating potential setup complexities. Some features like multi-modal model integration and workflow export are listed in the TODO list, suggesting ongoing development.