Google’s Agent Development Kit (ADK) for Java has taken a big step forward with its 0.2.0 release. This update introduces integration with the LangChain4j framework, giving developers access to a wider range of Large Language Models (LLMs), including third-party options.
Takeaways 🚀
✅LangChain4j expands ADK flexibility.
✅Supports third-party LLM models.
✅Simplifies AI agent development.
✅Improves tools and memory.
✅Enables local model integration.
Accessing Third-Party Models Made Easy
ADK for Java already supported Google Gemini and Anthropic Claude. Now, with LangChain4j, developers can use models from providers like OpenAI, GitHub, and Mistral. They can also work with local open-weight models through tools like Ollama and Docker Model Runner. This flexibility allows developers to choose models that best fit their needs, whether hosted in the cloud or running locally.
How LangChain4j Works with ADK
The integration uses a new LangChain4j class, which extends ADK’s BaseLlm abstract class. This connection makes it easy to add LangChain4j-supported models to ADK agents. For example, developers can configure agents to use Gemma 3 with Docker Model Runner or Qwen 3 with Ollama.
Here’s how you can set up a chess coach agent using Gemma 3:
LlmAgent chessCoachAgent = LlmAgent.builder()
.name("chess-coach")
.description("Chess coach agent")
.model(new LangChain4j(dmrChatModel))
.instruction("""
You are a knowledgeable chess coach
who helps chess players train and sharpen their chess skills.
""")
.build();
This setup shows how simple it is to connect third-party models while keeping agents versatile and powerful.
What’s New in Version 0.2.0
The 0.2.0 release doesn’t stop at LangChain4j integration. It also brings several new features to improve the development process:
- Better Tooling: Developers can now create tools from object instances, not just static methods. Asynchronous support has also improved, making agents more responsive.
- Smarter Agent Logic: New features like chained callbacks and better memory management give developers more control over agent behavior.
- Simpler Builds: A parent POM and Maven Wrapper make the build process easier and more consistent.
These updates make it faster and easier to build AI agents with ADK.
A Big Step for AI Development
The LangChain4j integration makes ADK for Java more open and flexible. Developers can now build AI agents with a wide variety of LLMs, tailored to their specific needs. This update simplifies the process of integrating third-party models while expanding what’s possible with ADK.
To learn more, check out the official release notes or explore the ADK for Java documentation. Whether you’re new to ADK or an experienced developer, this release offers exciting opportunities to create innovative AI agents.
