Get Notified: Introduction to Agent Observability
For the best experience, we recommend attendees have some familiarity with Grafana Cloud and are building or planning to build AI agents in production.
Sign up to be notified
when we’re hosting an Introduction to Agent Observability workshop either online or in person near you.
By signing up for notifications, you agree to be emailed about event details and related product-level information.
As AI agents move into production, uptime alone can't tell you whether they're actually working. With Agent Observability in Grafana Cloud, you'll bring evaluation, alerting, and tracing into one workflow, so you can catch quality regressions in agent behavior before your customers do.
You'll start with a tour of Agent Observability's end-to-end workflow across a single conversation, then build your own quality and safety evaluators and run them against live traffic. You'll put those detectors through a pre-launch fire drill and wire up the alerting and guards that Agent Observability identifies that you need. Finally we'll work through a simulated production incident: a live campaign's pass rate drops, and you'll ship a prompt fix and trace the issue back to its root cause.
Highlights:
Tour of Agent Observability, from the agent handoffs and tool calls through to its evaluation scores
Guide to evaluations, matchers, and what sampling costs at scale
Learn about experiments and test suites for offline evaluation to test a prompt change before it reaches production
A pre-launch fire drill against your own detectors, covering prompt injection, toxicity, and bias
Live Q&A with Grafana Labs' experts
Workshops are not recorded for on-demand consumption post-event. Live attendance and participation in breakout sessions are required.