Observability Without Overs...
summary
Hands-on workshop shows a CLI chatbot + backend running a local Olama 54-mini model, wires telemetry through an OpenTelemetry collector and OpenLIT SDK, and exports metrics to Prometheus and traces to Tempo with a Grafana dashboard. Observability initially captures prompts including emails and names; a Presidio anonymizer intercepts and masks PII before instrumentation, preserving token/cost metrics while removing sensitive fields. Deployment uses GitHub Codespaces; Presidio downloads a ~500MB NLP model and adds modest latency. Tools: OpenTelemetry + OpenLIT, Presidio anonymizer, Prometheus + Tempo + Grafana.
AI-written summary · about
chapters (11)
- 0:00Welcome
- 1:50Setting up the hands-on workshop environment
- 4:05Observability for AI agents
- 6:14Privacy paradox
- 7:19Workshop overview
- 9:50Lab 1: Deploy AI agent
- 17:44Lab 2: Basic observability
- 24:28Lab 3: Prometheus & Grafana
- 30:32Lab 4: Privacy techniques
- 40:37Workshop recap
- 41:19Q&A
related talks
01h 02mObserving AI Applications with OpenLit and OpenTelemetry
37mObservability For You And Me With Opentelemetry
33mWhat Did My Agent Do? Observability and Accountability for AI Agents
53mUnmask hidden issues with Observability superpowers!
10mDeep Dive Into Observability With OpenTelemetry: A Hands-On Codelab
31mCharting New Frontiers: The Evolution of Observability in Cloud Environments
35mOpenTelemetry for Busy People
29mOTel You It's Not Just for Backend!