Tool use with LangChain4j and Ollama

This example shows LangChain4j tool calling against a local Ollama model.

A plain Java method annotated with @Tool becomes callable by the model, and LangChain4j runs the tool loop on behalf of the AI service. Small models call tools unreliably, so a larger model such as llama3.2 gives more consistent results.

Requirements

• Java 21+
• Maven 3.9+
• Ollama installed locally
• A pulled chat model such as llama3.2:1b

Steps for running this example in the shell

1. Start Ollama and pull the model:

ollama serve
ollama pull llama3.2:1b

2. Run the application:

mvn compile exec:java

Output

42