# [[Inside the Grafana AI Team Weekly 2026-05-19]] <iframe width="560" height="315" src="https://www.youtube.com/embed/P3AX1H527Gs" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe> plugins:: "" Related:: "" ## Description The Grafana AI team shares what's new in AI Observability this week: a new way to instrument and visualize agent workflows, plus a neat trick for jumping straight from a metric spike to the exact conversation that caused it using Prometheus exemplars. In this episode: - Instrument and visualize workflows built with frameworks like LangGraph — see each step, its loops, LLM calls, cost, and input/output state, so you can spot exactly where an agent run failed. - A redesigned agents feed with evaluation/quality data and the ability to select multiple conversations into a collection. - Using Prometheus exemplars in AI Observability to attach trace IDs to metrics (like time to first token) and jump from a latency spike to the real trace behind it — instrumented for free via Grafana Assistant. - A look at the Sigil SDK plugins, including tracking your Claude Code usage. This is part of our weekly series where the Grafana AI team demos work in progress on AI Observability and related tooling. #Grafana #AIObservability #OpenTelemetry #LLM #Observability ## Chapters ``` 00:00 What's new in AI Observability this week 00:39 Demo: Instrumenting & visualizing agent workflows (LangGraph) 02:39 Demo: Prometheus exemplars — metric spike to conversation 04:56 Tracking Claude Code usage with the Sigil SDK ``` ## Links and resources AI Observability and workflows: - [Grafana AI Observability overview](https://grafana.com/docs/grafana-cloud/machine-learning/ai-observability/) - [Instrument agents with frameworks (LangGraph, LangChain, etc.)](https://grafana.com/docs/grafana-cloud/machine-learning/ai-observability/guides/instrument-agents/) - [GenAI agent observability](https://grafana.com/docs/grafana-cloud/monitor-applications/ai-observability/genai/agent-observability/) - [Get started with AI Observability](https://grafana.com/docs/grafana-cloud/machine-learning/ai-observability/get-started/) Exemplars: - [Introduction to exemplars](https://grafana.com/docs/grafana/latest/fundamentals/exemplars/) - [Configure and use exemplars (Grafana Cloud)](https://grafana.com/docs/grafana-cloud/send-data/traces/configure/exemplars/) > [!note] To verify before publishing > "Sigil SDK" is the internal name and likely has no public doc under that name — externally it maps to the AI Observability SDKs (see the instrument-agents link above). Confirm the exact public repo/link before publishing. ## Transcript ![[Inside the AI Team Weekly (May 19th 2026)_transcript]]