ConceptLLM可观测性 / AI可观测性

LLM Observability

针对大语言模型应用的可观测性细分领域，关注AI系统的语义质量监控、输出漂移检测和静默失败识别，代表工具包括LangSmith、Helicone、Arize AI等

Timeline (last 90 days)

Oct 15

LLM observability is the adaptation of the three pillars of traditional software observability—logs, metrics, and traces—to pipelines centered on LLMs.

Unverified50%

All Facts (1)

Unverified

LLM observability is the adaptation of the three pillars of traditional software observability—logs, metrics, and traces—to pipelines centered on LLMs.

50%

Source Articles

Dependency Drift in Production AI Agents: Detection, Response, and Building Resilient SystemsJul 22