Observability Sessions Stockholm

AI is changing how your team writes code, ships changes, and investigates incidents; raising the bar on reliability, governance, and trust. If you lead engineering, platform, SRE, or observability, the question is: how do you move faster with AI without losing control?This session covers five pillars for an AI-first observability strategy: tying AI and observability to business outcomes, building trust through evaluation and evidence, getting your platform and data foundation right, adopting AI safely through governance and culture, and building a roadmap from AI-assisted workflows to autonomous operations.We will ground the discussion in practical lessons from our own experience at Grafana Labs, as well as from the customers we are helping navigate these changes. We will share what we are learning as a software company building with AI, as an engineering organization operating reliable systems, and as an observability provider helping teams move from insight to action across the software lifecycle.

With more than 1,780 stores, NorgesGruppen operates technology at a scale where traditional monitoring quickly becomes difficult to manage. In our previous setup, we were dealing with around 12,000 alerts every month. The problem wasn't a lack of data or alarms: we simply couldn't see the forest for the trees.As part of our Empower Our People initiative, we set out to change how we operate our store technology: moving from reactive support to proactive observability. A key principle has been simple: if we can write a runbook for a known problem, we should be able to automate it. Today, a growing share of known issues can be detected and handled automatically, allowing our teams to spend less time reacting to alarms and more time improving the customer and store experience.Our Computer Vision project at self-checkout takes this thinking one step further. The solution is currently deployed in 10 stores, with an ambition to cover around 2,000 checkout lanes. The goal is simple: reduce friction for customers when selecting the right product at self-checkout. Our model can identify the correct product with 99% accuracy, making the checkout experience faster and easier for the customer.And we don't just deploy the model and hope for the best. We continuously monitor how it performs through what we call the "Banana Calculator," giving us a practical way to understand whether the technology is actually delivering the accuracy and frictionless experience we want in our stores.In this session, we'll share how NorgesGruppen moved from 12,000 alerts a month and reactive support towards proactive, increasingly automated operations, and how we're now applying the same observability mindset to Computer Vision at grocery scale.

For years, monitoring meant people building views, writing queries, and watching screens until something looked wrong.  This approach no longer keeps pace with the growing speed and complexity that come with agentic operations and ties up your most experienced engineers. This session shows what monitoring with AI looks like today. 

With Grafana Assistant, your team can ask questions in natural language and get straight answers, build what they need just by describing it, and find the root cause in minutes instead of hours. Problems get flagged before anyone notices them, and data costs stay in check without anyone worrying about the bill. The payoff: every engineer works like your most experienced, issues resolve faster, and observability becomes a source of better decisions, not just another cost. 

A candid conversation with observability leaders on what it really takes to build and sustain a strong practice. We'll dig into the organizational challenges that tools alone don't solve: getting executive buy-in, breaking down silos between Dev, Ops, and SRE, controlling cost as data grows, proving the value of observability to the business, and the cultural shifts that separate teams that mature from teams that stall. Real lessons, honest trade-offs, and audience questions welcome.

LLMs and agents are changing how software behaves. Modern AI applications do more than generate text. They call tools, retrieve context, coordinate across systems, and power business-critical applications. That creates a new production risk: a response can pass every technical check and still be wrong. Healthy isn't the same as correct and traditional observability was never built to tell the difference.

This session shows why AI observability is now essential for teams building LLM-powered applications or agents. You'll see what AI changes, then how Agent Observability helps you control cost and performance, debug agents faster, protect user trust, and ship better agents through evaluation and production feedback.

We'll close with a demo on an enterprise use case: an AI agent that handles user requests, calls tools, and processes sensitive data and how you’d monitor, evaluate, and improve like any production system.

Hear directly from the Grafana Labs VP of Product on the future of observability and how AI-driven insights and tools such as Adaptive Telemetry lead to better data-driven decision making.