SmartGraph makes it easy to integrate powerful AI assistants into your applications using the CompletionComponent. This component allows you to leverage large language models (LLMs) like GPT-4 or Claude to create intelligent, context-aware applications. In this guide, we’ll explore how to effectively use the CompletionComponent to build sophisticated AI assistants.

​
Basic Usage

To get started with AI assistants in SmartGraph, you’ll use the CompletionComponent. Here’s a simple example:

import asyncio
import os
from dotenv import load_dotenv
from smartgraph.components import CompletionComponent
from smartgraph.logging import SmartGraphLogger

# Load environment variables
load_dotenv()

# Set up logging
logger = SmartGraphLogger.get_logger()
logger.set_level("INFO")

async def main():
    # Create a CompletionComponent
    completion = CompletionComponent(
        "AI_Completion",
        model="claude-3-haiku-20240307",
        temperature=0.7,
        api_key=os.getenv("ANTHROPIC_API_KEY"),
    )

    user_input = "Explain reactive programming"
    logger.info(f"Processing input: {user_input}")

    try:
        result = await completion.process({"message": user_input})

        if "error" in result:
            logger.error(f"Error during processing: {result['error']}")
            print(f"\nAn error occurred: {result['error']}\n")
        else:
            logger.info(f"Received: {result}")
            print(f"\nAI Response: {result.get('ai_response', 'No response')}\n")

    except Exception as e:
        logger.error(f"Unexpected error during processing: {str(e)}", exc_info=True)
        print(f"\nAn unexpected error occurred: {str(e)}\n")

if __name__ == "__main__":
    asyncio.run(main())

In this example, we create a simple AI assistant that can answer questions about reactive programming.

​
Customizing the AI Assistant

The CompletionComponent offers several customization options:

​
Model Selection

Choose the AI model you want to use:

CompletionComponent(
    "AI_Completion",
    model="claude-3-haiku-20240307",
    api_key=os.getenv("ANTHROPIC_API_KEY"),
)

​
Temperature

Control the randomness of the AI’s responses:

CompletionComponent(
    "CreativeAssistant",
    model="claude-3-haiku-20240307",
    temperature=0.8,  # Higher temperature for more creative responses
    api_key=os.getenv("ANTHROPIC_API_KEY"),
)

​
Streaming Responses

Enable streaming for real-time responses:

CompletionComponent(
    "StreamingAssistant",
    model="claude-3-haiku-20240307",
    stream=True,
    api_key=os.getenv("ANTHROPIC_API_KEY"),
)

​
Advanced Usage

​
Integrating External Knowledge

Enhance your AI assistant with external knowledge using the DuckMemoryToolkit:

import asyncio
import os
from dotenv import load_dotenv
from smartgraph.components import CompletionComponent
from smartgraph.tools.duck_memory_toolkit import DuckMemoryToolkit
from smartgraph.logging import SmartGraphLogger

# Load environment variables
load_dotenv()

# Set up logging
logger = SmartGraphLogger.get_logger()
logger.set_level("INFO")

async def main():
    # Initialize DuckMemoryToolkit
    memory_toolkit = DuckMemoryToolkit("assistant_memory.duckdb")

    # Create a CompletionComponent with DuckMemoryToolkit
    completion = CompletionComponent(
        "AI_Completion",
        model="claude-3-haiku-20240307",
        temperature=0.7,
        api_key=os.getenv("ANTHROPIC_API_KEY"),
        toolkits=[memory_toolkit],
    )

    # Add some initial knowledge
    await memory_toolkit.add_memory("python", {"language": "Python", "type": "Programming language", "creator": "Guido van Rossum"})

    user_input = "Tell me about Python and its creator"
    logger.info(f"Processing input: {user_input}")

    try:
        result = await completion.process({"message": user_input})

        if "error" in result:
            logger.error(f"Error during processing: {result['error']}")
            print(f"\nAn error occurred: {result['error']}\n")
        else:
            logger.info(f"Received: {result}")
            print(f"\nAI Response: {result.get('ai_response', 'No response')}\n")

        # Store the interaction in memory
        await memory_toolkit.add_memory(f"interaction_{user_input}", {"question": user_input, "answer": result.get('ai_response', 'No response')})

    except Exception as e:
        logger.error(f"Unexpected error during processing: {str(e)}", exc_info=True)
        print(f"\nAn unexpected error occurred: {str(e)}\n")

if __name__ == "__main__":
    asyncio.run(main())

This setup allows the AI assistant to access and store information using the DuckMemoryToolkit, providing a way to maintain knowledge across sessions.

​
Best Practices

  1. API Key Security: Always use environment variables or secure key management systems for API keys.

  2. Error Handling: Implement robust error handling to manage potential issues with the AI service.

  3. Logging: Use the SmartGraphLogger for consistent logging across your application.

  4. Memory Management: When using memory toolkits, consider implementing a strategy for managing the growth of stored data.

  5. Rate Limiting: Implement rate limiting to avoid exceeding API quotas and manage costs.

​
Conclusion

The CompletionComponent in SmartGraph provides a powerful way to integrate AI assistants into your applications. By leveraging its features and following best practices, you can create sophisticated, context-aware AI applications that enhance user experiences and automate complex tasks.

​
Next Steps

Now that you understand how to integrate AI assistants using the CompletionComponent and enhance them with external knowledge, explore how to create custom components in the Custom Components section to further extend your SmartGraph applications.