Grafana Labs

⏳The clock is ticking! Last day to save 30% on ObservabilityCON 2026 tickets. Whether you're building in a local environment or operating agentic workflows in production, ObservabilityCON brings together all the latest innovations, thought leaders, and experts to help you (and your agents) navigate the fast-moving AI era. Don’t miss the hands-on workshops, the live keynote, the technical deep-dive sessions, and talks from Anthropic, Deutsche Telekom, Uber, LexisNexis Risk Solutions, UKG, and more. Register: https://lnkd.in/gG3B39wE

👋 AI for observability. Observability for AI. Learn more at our booth at Observability Summit EU. Come get some swag and answers to your observability questions. And don’t miss Imma Valls Bernaus’s talk.

🤔 Just *how* powerful is the Grafana Knowledge Graph? See for yourself. Jia Xu demos how it powers root-cause investigations both in the Grafana UI and through agentic workflows using gcx and Claude Code. Give it a watch and let us know what you think: https://lnkd.in/g_RcYFeD

After migrating to Grafana Cloud, Pintu's incident response costs went down by 38%! 🤯 Pintu, one of the top three crypto exchanges in Indonesia, runs 24/7 trading operations where every incident hits revenue and user trust directly. The impact of migrating to Grafana Cloud in a nutshell: • Migrated 112 users to Grafana IRM in four months — with no major incident • Cut incident-response costs by roughly 38% • Consolidated eight self-managed Grafana instances into a single Grafana Cloud account • Replaced manual schedule changes with self-service swaps via Grafana Cloud RBAC • Built a single page showing who's on call for every team, one click from paging them through Grafana IRM Watch/read Pintu's case study: https://lnkd.in/gHsA7XUB

ICYMI: AI companies like 7AI, TeamSystem, and Zama have standardized on Grafana Cloud as AI workloads generate more telemetry. Here's why 👇

🤖 Our engineers started applying SLO thinking to agent behavior: Give each behavior its own evaluation (groundedness, fulfillment, toxicity, cost), set a target, and now you have an error budget for hallucinations instead of just crossing your fingers and hoping for the best. The post walks through how this works in Grafana Cloud's Agent Observability, including how deterministic checks and model-based judges sit side by side, and how Grafana Assistant can spin up evaluations against real conversations: https://lnkd.in/gXbWABbM

🧡 Souls intact. Grafana Labs CRO Dave Kranowitz shares how he thinks about this guiding principle alongside success within the Sales team. We want to be able to look each other in the eye decades from now and respect not only the work we did together, but *how* we accomplished it. If this resonates with you, we're hiring. Check out our openings and apply: https://lnkd.in/dW6a4JAH

Running LLMs as coding agents without observability is almost like flying with your eyes closed. 🙈 The gcx CLI tool and Grafana MCP server bring telemetry-driven development to AI workflows, so you can actually see what your agents are doing, catch unexpected behavior, and build with confidence.

Still playing Wordle? We put k6 agentic testing to the... test, and it did a pretty decent job with it!

With AI moving so quickly, it can be tempting to chase each and every development, but that’s not how we work at Grafana Labs. 🏃♀️🏃♀️🏃♀️ Of course, we test new models as they arrive to see how they fit our workflows, but the models themselves are hardly a complete solution. The real work is understanding problems, designing experiences, building the right systems around them, and making sure the experiences earn people’s trust. As the AI team has grown over the last year, this is more important than ever: we will never ship AI slop tools, or AI for the sake of it. To get this right, every piece of your feedback, good and bad, helps us. It ensures we’re solving real problems: https://lnkd.in/gm4b6EbQ