AI has changed who ships software and how fast. New services hit production every week, often built by engineers who have never had to think hard about observability, and SLO creation hasn't kept up. Finding the right metric, writing a query that doesn't silently return bad data, and picking a defensible target still takes days of work. Most teams end up rebuilding the same SLIs again and again.
Brian Singer and Andrzej Voss will show two things we've built to close that gap. The first is easySLO, with pre-built SLIs for APIs, Kubernetes workloads, latency, availability, and LLM-based services. You can stand up a production-grade SLO from telemetry you already have without writing a single query, and SLI Analyzer reads your history to recommend a target you can actually defend.
The second is for the moment your error budget starts burning. Nobl9 knows an SLO is burning, but the cause lives somewhere inside your infrastructure. The alert explainer connects to your own observability stack, investigates the alert against your traces and metrics, and writes what it finds back into Nobl9 as an annotation. It can narrow a burn down to a specific request type, user, or oversized response. Sometimes it concludes an alert is simply noisy and needs no action, which is also worth knowing.
