Introduction
Before you instrument anything, it helps to understand the problem agentic systems create, what Splunk Agent Observability gives you to solve it, and the application you’ll be working with for the rest of the workshop.
Persona
You’re an AI engineer at Careful Health Provider. Your team is launching an agentic AI
healthcare assistant that answers patient questions about medications and looks up patient
records. Leadership loves the demo, but they’re nervous about what happens in production,
where a single hallucinated dosage or leaked record can become a safety incident and a
headline. Your job: make the agent’s behavior observable, measurable, and safe before and
after it ships.
What makes agentic AI different
You already have great observability for infrastructure and application performance. Agentic AI is different: the failures are about reasoning, quality, and safety, not just latency and errors. This workshop gives you the tools to see and govern that new layer.
Continue to the subsections to understand the challenge, the key concepts, and the application.