In the ever-evolving world of AI applications, ensuring optimal performance is paramount. With Graphsignal, developers gain a powerful observability platform specifically designed for AI agents and applications powered by large language models (LLMs). This article will guide you through the setup and integration of Graphsignal, making it straightforward and user-friendly.

What Graphsignal Offers

• Trace generations, runs, and sessions with complete AI context.
• Score user interactions and application executions for performance tracking.
• Analyze latency breakdowns and distribution metrics.
• Monitor API, compute, and GPU utilization effectively.
• Receive notifications regarding any errors or anomalies.
• Evaluate model API costs across deployments, models, or users.

With these features, Graphsignal ensures you can maintain high-quality AI applications that provide exceptional user experiences.

Installation Steps

To get started with Graphsignal, follow these simple installation steps:

• Open your terminal and run the following command to install the Graphsignal library:

pip install --upgrade graphsignal

Configuration

Once installed, you need to configure the Graphsignal tracer:

• Import the Graphsignal module into your Python code:

import graphsignal

• Specify your API key directly in your code or use the GRAPHSIGNAL_API_KEY environment variable:

graphsignal.configure(api_key='my-api-key', deployment='my-app')

• To obtain your API key, sign up for a free account at graphsignal.com.
• You can also configure the tracer via the command line:

python -m graphsignal script
python -m graphsignal -m module

Integration

Now, it’s time to integrate Graphsignal with your existing libraries and frameworks. Graphsignal auto-instruments libraries like OpenAI and LangChain, capturing essential traces and data:

• Integrate with OpenAI and LangChain by following the guides provided:
• OpenAI Integration
• LangChain Integration
• Available guides for tracking include:
• Session Tracking
• User Tracking
• Manual Tracing
• Scores and Feedback
• Cost and Usage Monitoring

Analyze Your Applications

After configuration and integration, you can start monitoring your applications. Log in to Graphsignal to analyze performance and detect potential issues.

Understanding Overhead

A great advantage of the Graphsignal tracer is its lightweight nature. The overhead for each trace remains under 100 microseconds, ensuring minimal disruption to application performance.

Security and Privacy

Graphsignal prioritizes your data’s security. The tracer makes only outbound connections to api.graphsignal.com and cannot accept inbound connections or commands. Payloads such as prompts and completions are recorded by default, but you can disable this behavior by setting record_payloads=False in your configuration.

Troubleshooting

If you encounter issues, enable debug logging by adding debug_mode=True in your configuration. Should the debug log not resolve your problem, please reach out to the support team via your account.

Ensure outgoing connections to https://api.graphsignal.com are permitted in case of connection issues. For more insights, updates, or to collaborate on AI development projects, stay connected with fxis.ai.

At fxis.ai, we believe that such advancements are crucial for the future of AI, as they enable more comprehensive and effective solutions. Our team is continually exploring new methodologies to push the envelope in artificial intelligence, ensuring that our clients benefit from the latest technological innovations.