Atlas · GenAI 2026
LLM Observability
LLM observability & tracing (LangSmith, Phoenix, Datadog LLM)
conceptPeak: 2025Observability & TracingAI consensus: 3/3
Prerequisites
- mediumLLM Inference Serving
LLM observability tools monitor inference engines — understanding what the engine does helps interpret the traces
- mediumAI Agent Design
Tracing agent workflows (tool calls, reasoning steps) is a primary LLM observability use case
Recommended reference
docs.smith.langchain.com — LangSmith docs; plus arize.com/docs-phoenix for open-source alternative
Notes from AI deep research
Anthropic Opus
LangSmith, Phoenix, Datadog LLM. TW Radar: az 4 narzedzia! Dyscyplina dojrzewa blyszkawicznie
OpenAI Deep Research
Prompt versioning, token cost tracking [OA#48]
Google Deep Think
Pełne mapowanie drzewa wywołań [G#76]
Related skills
- → is subcategory of: MLOps(3/3)
- → is subcategory of: ML Monitoring(3/3)
- ← is part of: Stochastic System Debugging(3/3)
- → is part of: LLM Inference Serving(2/3)
- ← is part of: LLM Evaluation Frameworks(1/3)
- ← is an instance of: LangSmith(0/3)
- ← is an instance of: OpenTelemetry(0/3)