alibaba/spring-ai-alibaba View GitHub Homepage for Latest Official Releases

Enterprise-grade agentic AI framework for Java developers, built on Spring AI with deep Alibaba Cloud integration for building intelligent agents, workflows, and multi-agent applications

Apache-2.0Javaspring-ai-alibabaalibaba 7.5k Last Updated: December 15, 2025

Spring AI Alibaba - Agentic AI Framework for Java Developers

Overview

Spring AI Alibaba is an enterprise-grade AI application development framework specifically designed for Java developers. Built on top of Spring AI and deeply integrated with Alibaba Cloud's Bailian platform, it provides a comprehensive solution for building intelligent agents, workflows, and multi-agent applications in production environments.

Official Repository: https://github.com/alibaba/spring-ai-alibaba
Official Website: https://java2ai.com
License: Open Source
Requirements: JDK 17+

Core Architecture

The framework consists of three fundamental components:

1. Agent Framework

A ReactAgent-based development framework centered around the ReAct (Reasoning + Acting) paradigm. It enables developers to build intelligent agents with automatic context engineering and human-in-the-loop capabilities. For complex scenarios, it provides built-in workflow patterns including:

  • SequentialAgent: Execute agents in sequential order
  • ParallelAgent: Run multiple agents concurrently
  • RoutingAgent: Route requests based on conditions
  • LoopAgent: Implement iterative workflows

2. Graph Runtime

A low-level workflow and multi-agent orchestration framework inspired by LangGraph. It features:

  • Rich set of prebuilt workflow nodes
  • Simplified Graph State definitions
  • Native streaming support
  • Human-in-the-loop integration
  • Memory and persistent storage
  • Graph state snapshots
  • Nested and parallel graph execution
  • PlantUML and Mermaid format export

3. Spring Boot Starters

Integration starters that connect Agent Framework with enterprise services like Nacos, providing:

  • Agent-to-Agent (A2A) communication
  • Dynamic configuration management
  • Distributed MCP discovery and routing

Key Features

Intelligent Agent Development

  • ReactAgent Pattern: Build agents with reasoning and acting capabilities following the ReAct paradigm
  • Multi-Agent Orchestration: Compose multiple agents for complex task execution
  • Context Engineering: Built-in best practices for prompt engineering and context management
  • Human In The Loop: Seamlessly integrate human feedback and approval steps
  • Streaming Support: Real-time streaming of agent responses with token-level granularity
  • Error Handling: Robust error recovery and retry mechanisms

Model & Tool Integration

  • Multiple LLM Providers: Support for DashScope (Qwen, DeepSeek), OpenAI, and other providers
  • Tool Calling: Comprehensive function calling capabilities with FunctionToolCallback
  • Model Context Protocol (MCP): Full MCP support for standardized model interactions
  • Structured Output: Define output schemas and types for format control
  • Multimodal Support: Handle text, images, and other modalities

Enterprise-Ready Features

  • Nacos MCP Registry: Distributed MCP Server discovery and load balancing
  • Higress AI Gateway: LLM model proxy and routing
  • ARMS Observability: Integration with Alibaba Cloud Application Real-Time Monitoring Service
  • Langfuse Integration: Comprehensive tracing and evaluation
  • Vector Stores: Support for multiple vector databases (Hologres, AnalyticDB, OpenSearch)
  • RAG Support: Complete retrieval-augmented generation pipeline
  • Chat Memory: Session memory management for multi-turn conversations
  • NL2SQL: Natural language to SQL transformation

Additional Capabilities

  • Document Parsing: Support for multiple formats (PDF, Word, Excel, etc.)
  • Image Generation: DashScope-based image model integration
  • Audio Processing: Audio transcription and synthesis
  • Prompt Management: Dynamic prompt templates with Nacos integration
  • Workflow Visualization: Export workflows to PlantUML and Mermaid formats

Getting Started

Quick Start Example

Add dependencies to your Spring Boot project:

<dependencyManagement>
  <dependencies>
    <dependency>
      <groupId>com.alibaba.cloud.ai</groupId>
      <artifactId>spring-ai-alibaba-bom</artifactId>
      <version>1.1.0.0-M5</version>
      <type>pom</type>
      <scope>import</scope>
    </dependency>
  </dependencies>
</dependencyManagement>

<dependencies>
  <dependency>
    <groupId>com.alibaba.cloud.ai</groupId>
    <artifactId>spring-ai-alibaba-agent-framework</artifactId>
  </dependency>
  <dependency>
    <groupId>com.alibaba.cloud.ai</groupId>
    <artifactId>spring-ai-alibaba-starter-dashscope</artifactId>
  </dependency>
</dependencies>

Simple ReactAgent Example

// Initialize ChatModel
DashScopeApi dashScopeApi = DashScopeApi.builder()
    .apiKey(System.getenv("AI_DASHSCOPE_API_KEY"))
    .build();

DashScopeChatModel chatModel = DashScopeChatModel.builder()
    .dashScopeApi(dashScopeApi)
    .build();

// Create ReactAgent
ReactAgent writerAgent = ReactAgent.builder()
    .name("writer_agent")
    .model(chatModel)
    .description("A professional writer agent")
    .instruction("You are a renowned writer skilled in creative writing.")
    .outputKey("article")
    .build();

// Call the agent
AssistantMessage message = writerAgent.call("Write a 100-word essay about AI");

Multi-Agent Workflow Example

// Create reviewer agent
ReactAgent reviewerAgent = ReactAgent.builder()
    .name("reviewer_agent")
    .model(chatModel)
    .description("Reviews and edits articles")
    .instruction("You are an expert editor who reviews and improves content.")
    .outputKey("reviewed_article")
    .build();

// Compose agents in sequence
SequentialAgent blogAgent = SequentialAgent.builder()
    .name("blog_agent")
    .description("Writes and reviews articles")
    .subAgents(List.of(writerAgent, reviewerAgent))
    .build();

// Execute workflow
Optional<OverAllState> result = blogAgent.invoke("Write a blog post about Spring AI");

Official Products & Examples

JManus

A Java implementation of Manus (general AI agent) built with Spring AI Alibaba. It supports:

  • Autonomous planning and execution
  • Fine-tuned agents for specific business scenarios
  • Customized tools and sub-agents
  • Plan adjustment and reuse capabilities
  • Currently used in many applications within Alibaba Group

DeepResearch

An intelligent research agent featuring:

  • Complete front-end web UI and backend implementation
  • Web search and crawling capabilities
  • Python script engine integration
  • MCP service support
  • Generates comprehensive research reports using LLMs and tools

DataAgent

A natural language to SQL project that enables:

  • Direct database queries using natural language
  • No need to write complex SQL statements
  • Seamless integration with enterprise databases

Playground

A comprehensive example application with:

  • Complete frontend UI and backend implementation
  • Demonstrates all core framework capabilities
  • Features: chatbot, multi-round conversations, image generation, multimodality, tool calling, MCP, RAG
  • Available for local deployment and customization

Spring AI Alibaba Admin

Local visualization toolkit providing:

  • Project management
  • Runtime visualization
  • Tracing and debugging
  • Agent evaluation tools

Ecosystem Integration

Alibaba Cloud Services

  • Bailian Platform: LLM model services (Qwen series, DeepSeek)
  • DashScope: Comprehensive AI model service platform
  • Vector Stores: Hologres, AnalyticDB, OpenSearch
  • ARMS: Application monitoring and observability
  • Nacos: Configuration and service discovery

Third-Party Integrations

  • Langfuse: Tracing and evaluation
  • OpenAI: Compatible with OpenAI API
  • Higress: AI gateway for model routing
  • Spring AI: Built on Spring AI core concepts

Version Information

Current Stable Version: 1.0.0.2 (GA)
Latest Version: 1.1.0.0-M5 (Milestone)

Version 1.1.x Features

  • Enhanced agent development modes (Agentic, Multi-agent, Workflow)
  • Improved Graph Runtime
  • Enhanced A2A communication
  • Better MCP integration
  • Upgraded documentation and official website

Version 1.0.x Features

  • First GA release
  • Production-ready framework
  • Complete ChatBot, Workflow, and Multi-agent support
  • Deep Alibaba Cloud integration
  • Graph-based multi-agent framework

Available Starters

  • spring-ai-alibaba-starter-dashscope - DashScope model integration
  • spring-ai-alibaba-agent-framework - Agent framework core
  • spring-ai-alibaba-graph-core - Graph runtime
  • spring-ai-alibaba-starter-nl2sql - Natural language to SQL
  • spring-ai-alibaba-starter-memory - Chat memory management
  • spring-ai-alibaba-starter-nacos-mcp-client - Nacos MCP client
  • spring-ai-alibaba-starter-nacos-mcp-server - Nacos MCP server
  • spring-ai-alibaba-starter-nacos-prompt - Prompt management
  • spring-ai-alibaba-starter-arms-observation - ARMS observability
  • spring-ai-alibaba-starter-rag - RAG capabilities

Community & Support

Communication Channels

  • DingTalk Group: Search 130240015687 and join
  • WeChat: Follow official WeChat public account
  • GitHub Discussions: Community discussions and Q&A
  • Official Documentation: https://java2ai.com

Related Repositories

Use Cases

Enterprise Applications

  • Customer service chatbots
  • Internal business automation
  • Document processing and analysis
  • Knowledge base Q&A systems
  • Code generation and review

Vertical Domain Agents

  • Domain-specific intelligent assistants
  • Automated research and reporting
  • Data analysis and visualization
  • Process automation and orchestration
  • Multi-step workflow execution

Development Tools

  • AI-powered programming assistants
  • Code review and optimization
  • Documentation generation
  • Testing and debugging support

Technical Advantages

  1. Java-Native: First-class support for Java ecosystem and Spring Boot
  2. Enterprise-Ready: Production-proven in Alibaba Group
  3. Cloud-Native: Deep integration with Alibaba Cloud infrastructure
  4. Flexible Architecture: Support for various development patterns (low-code, high-code, zero-code)
  5. Comprehensive Tooling: Rich set of tools and components for agent development
  6. Active Community: Strong community support and regular updates
  7. Best Practices: Built-in patterns from real-world enterprise deployments

Comparison with Other Frameworks

Unlike Python-based frameworks (LangChain, LangGraph), Spring AI Alibaba provides:

  • Native Java support with Spring Boot ecosystem integration
  • Enterprise-grade features (observability, gateway, configuration management)
  • Production-ready deployments with cloud-native infrastructure
  • Seamless integration with Alibaba Cloud services
  • Strong typing and compile-time safety
  • Superior performance in JVM environments

Future Roadmap

  • Continuous framework optimization and performance improvements
  • Enhanced agent development patterns and tools
  • Expanded model provider support
  • More comprehensive documentation and tutorials
  • Additional enterprise integration capabilities
  • Community-driven features and improvements

Conclusion

Spring AI Alibaba represents a significant advancement in Java-based AI application development, bringing enterprise-grade agent frameworks to the Java ecosystem. With its comprehensive features, deep cloud integration, and production-proven architecture, it enables Java developers to build sophisticated AI applications with the same productivity and reliability they expect from the Spring ecosystem.

Whether you're building simple chatbots, complex multi-agent systems, or enterprise workflow automation, Spring AI Alibaba provides the tools, patterns, and infrastructure needed to bring your AI applications from demo to production.

Star History Chart