RuoYi AI

// Overview

RuoYi AI is a full-stack, self-hosted AI assistant platform that arrives ready to run rather than as a library you assemble. It bundles an admin panel and a user-facing front end on top of a Spring Boot backend, so a team can stand up a branded assistant with accounts, knowledge bases and workflows without building the surrounding application first.

The backend is Spring Boot 3.5.8 with Langchain4j, storing state in MySQL 8.0 and Redis and vectors in Milvus, Weaviate or Qdrant; the front end is Vue 3 on Vben Admin with element-plus-x, and Sa-Token plus JWT handle security. Communication is over WebSocket and SSE, so streaming responses and live updates work out of the box, and the platform ships logging, performance monitoring and service health checks alongside the application itself.

Functionally it covers the four things an internal assistant platform usually needs: model management across several providers including DeepSeek, Zhipu, MIMO, Bailian and OpenAI with multimodal understanding and integrations for Coze, Dify, FastGPT and RAGFlow; knowledge management with local RAG over a vector database and document parsing for PDF, Word and Excel; tool management through MCP and a skills layer; and orchestration, both as a drag-and-drop visual workflow designer with SSE streaming execution and as a Langchain4j-based multi-agent framework with a Supervisor mode that can combine tools and skills across several decision models. The project is split across repositories — backend, user front end, admin panel and companion apps — each mirrored on GitHub, Gitee and GitCode.

// What it does

• Multi-provider model management (DeepSeek, Zhipu, MIMO, Bailian, OpenAI) with multimodal understanding
• Local RAG over Milvus, Weaviate or Qdrant, with PDF, Word and Excel document parsing
• Visual workflow designer with drag-and-drop node orchestration and SSE streaming execution
• Multi-agent framework on Langchain4j with Supervisor-mode orchestration across several decision models
• MCP protocol integration plus a skills layer for extending the tool ecosystem
• Platform integrations for Coze, Dify, FastGPT and RAGFlow
• Ships an admin panel and a user front end, with Sa-Token and JWT security and built-in logging and health checks

// Getting started

RuoYi AI is deployed with Docker Compose. The one-click path starts every service — backend, admin panel, user front end and their dependencies — from a single Compose file, using public GHCR images that need no docker login.

git clone --depth 1 --branch v3.1.0 https://github.com/ageerle/ruoyi-ai.git
cd ruoyi-ai

cp docs/docker/ruoyi-ai/.env.example docs/docker/ruoyi-ai/.env

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Stand up an internal, branded AI assistant with user accounts and an admin panel without writing the application shell
• Put a company knowledge base behind RAG on a self-hosted vector database rather than a third-party service
• Let non-developers assemble automations in a visual workflow designer that can call models, send email and pause for manual review
• Build agent features inside an existing Java and Spring Boot stack using Langchain4j instead of adding a Python service

// How RuoYi AI compares

RuoYi AI alongside other open-source low-code & no-code builders tools AI/TLDR tracks, ranked by GitHub stars.