Track your AI Agents FREE! (LangTrace)

Key Concepts

• AI Agents: Software entities designed to perform tasks autonomously.
• Observability: The ability to understand the internal state of a system from its external outputs.
• Langtrace: An open-source observability tool for AI applications.
• Langchain: A framework for building applications powered by language models.
• Llama Index: A data framework for building LLM applications.
• Crew AI: A framework for orchestrating multiple AI agents to work together.
• RAG (Retrieval-Augmented Generation): An AI framework that combines information retrieval with text generation.
• Docker: A platform for containerizing applications.
• API Key: A code used to authenticate and authorize access to an API.

1. Introduction to Langtrace

• AI agents are powerful, but optimizing them requires understanding their internal operations.
• Langtrace is introduced as an open-source observability tool designed to monitor AI applications.
• It allows users to track interactions, identify time-consuming functions, and optimize performance.

2. Key Features of Langtrace

• Metrics: Tracks total cost, human evaluation, dataset creation, prompts, registry, and settings.
• Backend Monitoring: Provides insights into backend processes that drive frontend chatbots.
• Integrations: Supports integrations with O Lama, Autogen, Gemini, Langchain, Llama Index, and Crew AI.
• Accessibility: Available on GitHub for free local use.

3. Installation Process Using Docker

• Step 1: Cloning the Repository:
  • Clone the Langtrace repository from GitHub using the command: git clone [repository URL] langtrace.
• Step 2: Docker Setup:
  • Ensure Docker is installed from docker.com.
  • Navigate to the Langtrace folder in the terminal.
  • Run docker compose up -d to start Langtrace.
• Step 3: Verification:
  • Verify the application is running using docker ps.
  • Access Langtrace at localhost:3000.
• Step 4: Login Credentials:
  • Find default credentials in the .env file within the cloned repository.
  • The default username is "admin" and the password can be found in the .env file.
  • Credentials can be modified in the .env file.

4. Setting Up Langtrace

• Project Creation:
  • Create a new project within the Langtrace application (e.g., "AI agents default").
• Integration Code:
  • Add two lines of code to integrate Langtrace into AI applications.
• API Key Generation:
  • Generate an API key within Langtrace and copy it.
• SDK Installation:
  • Install the Langtrace Python SDK using pip install langtrace llama-index langchain crewai.
• Environment Variables:
  • Export OpenAI API key: export OPENAI_API_KEY=[your_openai_api_key].
  • Export Langtrace API key: export LANTRACE_API_KEY=[your_langtrace_api_key].

5. Integration with Langchain

• Application Overview:
  • A simple RAG application that retrieves data from soccer rules and answers questions (e.g., "What is offside?").
• Integration Steps:
  • Import Langtrace: import langtrace.
  • Add the Langtrace API key and host to the environment variables.
• Running the Application:
  • Execute the Langchain application using python langchain_app.py.
• Monitoring:
  • Observe traces in the Langtrace UI, showing API calls and performance metrics.

6. Integration with Llama Index

• Application Overview:
  • Similar RAG application using Llama Index to search soccer rules and answer questions.
• Integration:
  • Integrate Langtrace using the same two lines of code as with Langchain.
• Execution:
  • Run the Llama Index application using python llama_index_app.py.
• Trace Analysis:
  • View detailed steps in Langtrace, such as document loading, embedding creation, and embedding retrieval.

7. Integration with Crew AI

• Setup:
  • Install required packages and export API keys.
  • Add the same two lines of Langtrace integration code.
• Monitoring:
  • Langtrace automatically tracks the Crew AI application.

8. Conclusion

• Langtrace provides valuable observability for AI applications, allowing developers to trace API calls and optimize performance.
• It supports multiple integrations and can be run locally for free.
• The presenter encourages viewers to try Langtrace and provide feedback.
• A recommendation is made to watch another video on creating RAG applications.