OpenTelemetry GenAI in Practice: What the Spec Says Vs. What You Actually See - Zach Groves, Datadog
Opentelemetry Genai Llm Monitoring Observability Telemetry Ai Instrumentation Semantic Conventions Cloud Native Kubecon Cncf
This talk dives into the practical implementation of OpenTelemetry's GenAI semantic conventions (v1.37+), comparing what the spec promises versus real-world adoption across different emitters like Strands, OpenLLMetry, and OpenInference. It is designed for DevOps engineers, platform teams, and developers building LLM-powered applications who need to instrument and monitor AI systems reliably.
Related videos
Beyond Dashboards: Cultivating the Observability Mindset - O. Hope Amaechi-Okorie, JSON Schema
OpenTelemetry Graduation (Grafana OTel Community Call #08)
The Speed of Metrics, the Fidelity of Traces: Architecting Post-Collection Aggregation - Zack Owens
Panel: Telemetry That Matters - Diana Todea, Antonio Jimenez Martinez & Laura Luttmer
Taming Tenancy, Cost and Architecture at Collibra Through OpenTelemetry and Our... Alex Van Boxel
DNS Tracing & Metrics Via eBPF in OpenTelemetry- Endre Sara, Causely & Nikola Grcevski, Grafana Labs
From Running to Operating: Real-World Strimzi in Production | Rajith Attapattu, Randoli
CNCF On-Demand: Introduction to OpenTelemetry Kotlin
The Invisible Tax: How Data Format Conversions Drive up Telemetry... Cijo Thomas & Joshua MacDonald
Keynote: 10 Million Spans Per Second: Lessons From Scaling OpenTelemetry at Reddit - Trevor Riles
Lightning Talk: Beyond Billions: Operating Thanos, Prometheus & OpenTelemetry... Narendra Sanikommu
