Arize Phoenix Observability
Trace, evaluate, and debug LLM and RAG applications with Arize Phoenix, an open-source AI observability platform. Capture OpenTelemetry spans, run evals on retrieval and responses, and find where your pipeline breaks.
This skill instruments LLM apps with Arize Phoenix. It sets up OpenTelemetry tracing to capture prompts, tool calls, and retrievals, runs evaluations for hallucination, relevance, and correctness, visualizes traces to pinpoint failures in RAG and agent pipelines, and compares runs to track quality over time.
arize-phoenix observability llm-eval opentelemetry tracing
When to use
Use when you need observability, tracing, or evaluation for an LLM, RAG, or agent application and want an open-source, OpenTelemetry-based tool.
Examples
Trace a RAG pipeline
Instrument retrieval and generation
Add Arize Phoenix OpenTelemetry tracing to my RAG app so I can see retrieved chunks and the final answer for each query
Evaluate response quality
Run hallucination and relevance evals
Set up Phoenix evals to score my LLM answers for hallucination and retrieval relevance across a test dataset