basic_llm

Basic LLM Example

This example demonstrates how to integrate Large Language Models (LLMs) with LangGraphGo using LangChain Go.

Overview

This example shows how to:

• Integrate OpenAI's LLM with LangGraphGo workflows
• Use LangChain Go for LLM interactions
• Handle message conversations in graph nodes
• Process AI-generated responses

Prerequisites

1. OpenAI API Key: Set your OpenAI API key as an environment variable:

   export OPENAI_API_KEY="your-api-key-here"
   

2. Dependencies: This example uses LangChain Go:

   go mod tidy
   

Features Demonstrated

LLM Integration

• Creating LangChain OpenAI client
• Configuring LLM parameters (temperature, etc.)
• Handling LLM responses within graph nodes

Message Handling

• Working with LangChain message formats
• Managing conversation context
• Appending AI responses to conversation history

Graph Workflow

• Single-node LLM processing
• State management with message arrays
• Error handling for LLM operations

Running the Example

cd examples/basic_llm
export OPENAI_API_KEY="your-api-key-here"
go run main.go

Code Structure

// Create LLM client
model, err := openai.New()

// Graph node with LLM processing
g.AddNode("generate", func(ctx context.Context, state interface{}) (interface{}, error) {
    messages := state.([]llms.MessageContent)
    
    // Generate response with LangChain
    response, err := model.GenerateContent(ctx, messages,
        llms.WithTemperature(0.7),
    )
    
    // Return updated conversation
    return append(messages, llms.TextParts("ai", response.Choices[0].Content)), nil
})

Expected Output

AI Response: 1 + 1 equals 2.

Key Concepts

• LangChain Integration: Using LangChain Go for LLM operations
• Message Management: Handling conversation context in graphs
• State Evolution: How conversation state grows through processing
• Error Handling: Managing LLM API errors in workflows

Extensions

You can extend this example by:

• Adding multiple LLM nodes for complex conversations
• Implementing different LLM providers
• Adding conversation memory and context management
• Integrating with streaming responses