Beyond Dashboards: Grafana Features & Secure AI Agents

🗒️ About the event
Join us at NLND Neuland Berlin for an evening of talks, pizza, drinks, and conversation. We’ll explore how to run coding agents securely and uncover Grafana features that can make observability and troubleshooting easier.

📍 NLND Neuland Berlin · Neuköllnische Allee 80, 12057 Berlin

📋 Agenda
18:00 · Welcome, pizza, and drinks 🍕
Grab a bite and meet fellow community members before the talks begin.
18:30 · Talks 🎙️

Grafana and the Chamber of Hidden Features
Grafana is best known for dashboards... but the Grafana ecosystem has grown far beyond panels, PromQL queries, and a few alert rules. In this session, we’ll explore the parts of Grafana that are easy to overlook: features hiding in plain sight, capabilities that don’t always make it into the typical Grafana workflow, and products and solutions from Grafana Labs that many teams may not have encountered yet. We’ll look beyond the obvious dashboarding use cases and uncover tools that can simplify observability, improve troubleshooting, and solve problems you might currently be addressing with entirely different systems.
Speaker: Anatolii Timoshuk
Anatolii Timoshuk used to run production systems. Now he talks to the people who run production systems — arguably a questionable career progression, but one that comes with fewer 3 a.m. alerts. After years in systems engineering and DevOps, building and operating large-scale cloud and Kubernetes infrastructure, Anatolii crossed over to the go-to-market side of tech. Today, as a Senior Observability Architect at Grafana Labs, he works with teams across EMEA and APAC on observability, architecture, and figuring out what's actually happening inside their systems. He brings an operator's mindset to Solutions Engineering: distrust the happy path, ask what happens when it breaks, and always look at the data.

So you want to run (coding) agents securely
Fixing a bug doesn’t need your cloud provider credentials. Summarizing a document doesn’t need permission to send emails. But useful agents need access to our code, data, and tools – and that can easily end up with more authority than the task requires. Unlike a malicious dependency, agents don’t necessarily need an attacker to start doing something you didn’t intend. “Just sandbox it” sounds like the obvious answer. So does “keep a human in the loop.” Neither quite settles what running an agent securely actually means. Tom explores the boundaries between isolation, autonomy, and trust – and how a perfectly reasonable task can end in what would be a felony if a human did it. What should an agent be allowed to reach? Which decisions should remain yours? And how would you notice when it starts going off track? Through live demos of popular open-source tools, Tom puts sandboxing to the test: what does it protect, what does it get in the way of, and what remains a matter of trust? Can you give an agent enough freedom to get the job done—and then just walk away?
Speaker: Tom Braack
Tom is a Senior Software Engineer at Grafana Labs, working on highly autonomous agents for automatic infrastructure observability and management.
Previously, Tom contributed to the OpenTelemetry Collector and Prometheus.
He draws on a strong cloud-native background, having previously created Grafana Tanka, an open-source tool for scalable Kubernetes configuration management.
Tom has studied Computer Science at FU Berlin and has presented about observability and cloud-native topics at conferences such as FOSDEM, PromCon, and OTel Community Day.

20:00 · Q&A and networking 💬
21:00 · Event ends 🔚
Whether you work in DevOps or site reliability engineering, or you’re just getting started with Grafana and observability, you’re welcome to join us. We look forward to seeing you there!

#Grafana #Observability