LLM Experiments: LLM Observability's Experiments

LLM Observability allows you to monitor, troubleshoot, and evaluate your LLM application.

• Agentic Workflow Monitoring: LLM Observability’s new graph-based visualization for monitoring agentic workflows allows you to troubleshoot agentic executions, including decisions, data flow, and hand-offs. To qualify for this Preview feature, your application should be written in Python and you should be using the OpenAI Agent SDK, LangGraph, or CrewAI.
• LLM Experiments: LLM Observability’s Datasets and Experiments allows you to build out ground truth dataset and run and compare experiments runs, so that you can confidently roll out to production after LLM, prompt, or code changes. To qualify for this Preview feature, your application should be written in Python.